<?page no="0"?> ISBN 978-3-381-15371-8 Gerhard Budin is a leading authority in terminology science, translation studies, digital humanities, and the philosophy of science. His influential work has shaped research agendas, advanced international standards, and fostered interdisciplinary collaboration. This volume celebrates his outstanding contributions and honors his 65 th birthday by bringing together scholars from terminology science, translation studies, and (computational) linguistics. Budin has played a key role in carrying the rich Viennese tradition of terminology science into the modern era through critical reflection and innovation. To honor this legacy, the volume includes original chapters that revisit fundamental research questions and showcase terminology work. His impact on the European language technology landscape is reflected in contributions on language representation and technology, while his linguistic background informs chapters on intertextuality and related topics. Translation forms a third major pillar, honored with original research on translation competence, technology, history, and localization. Gromann / Heinisch / Lušicky (Hrsg.) Beyond Terminology Dagmar Gromann / Barbara Heinisch / Vesna Lušicky (Hrsg.) Beyond Terminology A tribute to Gerhard Budin <?page no="1"?> Beyond Terminology <?page no="3"?> Dagmar Gromann / Barbara Heinisch / Vesna Lušicky (Hrsg.) Beyond Terminology A tribute to Gerhard Budin <?page no="4"?> DOI: https: / / doi.org/ 10.24053/ 9783381153725 © 2026 · Narr Francke Attempto Verlag GmbH + Co. KG Dischingerweg 5 · D-72070 Tübingen Das Werk einschließlich aller seiner Teile ist urheberrechtlich geschützt. Jede Verwertung außerhalb der engen Grenzen des Urheberrechtsgesetzes ist ohne Zustimmung des Verlages unzulässig und strafbar. Das gilt insbesondere für Vervielfältigungen, Überset‐ zungen, Mikroverfilmungen und die Einspeicherung und Verarbeitung in elektronischen Systemen. Alle Informationen in diesem Buch wurden mit großer Sorgfalt erstellt. Fehler können dennoch nicht völlig ausgeschlossen werden. Weder Verlag noch Autor: innen oder Herausgeber: innen übernehmen deshalb eine Gewährleistung für die Korrektheit des Inhaltes und haften nicht für fehlerhafte Angaben und deren Folgen. Diese Publikation enthält gegebenenfalls Links zu externen Inhalten Dritter, auf die weder Verlag noch Autor: innen oder Herausgeber: innen Einfluss haben. Für die Inhalte der verlinkten Seiten sind stets die jeweiligen Anbieter oder Betreibenden der Seiten verantwortlich. Internet: www.narr.de eMail: info@narr.de Druck: Elanders Waiblingen GmbH ISBN 978-3-381-15371-8 (Print) ISBN 978-3-381-15372-5 (ePDF) ISBN 978-3-381-15373-2 (ePub) Bibliografische Information der Deutschen Nationalbibliothek Die Deutsche Nationalbibliothek verzeichnet diese Publikation in der Deutschen Natio‐ nalbibliografie; detaillierte bibliografische Daten sind im Internet über http: / / dnb.dnb.de abrufbar. <?page no="5"?> 7 9 11 13 23 33 47 63 77 91 105 121 Inhalt Preface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Mira Kadrić Worte der Wertschätzung . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . RaDT Glückwunsch des Rats für Deutschsprachige Terminologie (RaDT) . . . . . . Christian Galinski Erinnerungen an 15 Jahre intensiver Zusammenarbeit mit Prof. Dr. Gerhard Budin. 1987~2001 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Dagmar Gromann, Barbara Heinisch & Vesna Lušicky Beyond Terminology: A tribute to Gerhard Budin . . . . . . . . . . . . . . . . . . . . . Klaus-Dirk Schmitz Begriffsbeziehungen. Umsetzung in Termbanken und Austauschformaten Rute Costa Terminological and Specialised Lexicographic Definitions: Navigating Connotation and Denotation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Georg Löckinger, Hendrik J. Kockaert Intensional definitions and concept modelling . . . . . . . . . . . . . . . . . . . . . . . . Tanja Wissik Research Infrastructures and Terminology: A Sustainability Perspective . . Juan Carlos Diaz Vásquez, María Cecilia Plested Alvarez Terminology work, principles, and applications: Decades of development from Colombia . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Elena Chiocchetti, Natascia Ralli Einflussfaktoren bei der Terminologiearbeit im Recht . . . . . . . . . . . . . . . . . . Frieda Steurs Terminology at the Dutch Language Institute: Towards a centre of expertise for professional language . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . <?page no="6"?> 135 155 171 187 195 211 225 245 263 Werner Winiwarter A Constructionist Pictorial Approach Towards a Universal Language Representation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Dragoș Ciobanu So much technology, so little time . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Jan Engberg Wie tief tauchen Vermittelnde in den Großbereich „Künstliche Intelligenz“ ein? Vorschlag einer Messung . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Wolfgang Dressler Text- und Diskursverkettung: Ein Beitrag zur Intertextualität . . . . . . . . . . . Peter Sandrini Organizational Translation Competence: Reframing Expertise . . . . . . . . . . Larisa Schippel Der Akademie-Verlag Berlin und sein translatorisches Verlagsprofil . . . . . Waltraud Kolb The automatic translation of literature and literary post-editing: German versions of a short story by Kate Chopin . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Xiaochun Zhang Multilingualism and Video Game Localisation . . . . . . . . . . . . . . . . . . . . . . . . Contributors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6 Inhalt <?page no="7"?> Preface For more than four decades, Gerhard Budin has shaped the landscape of terminology science, translation studies, multilingual knowledge organisation and digital humanities with a commitment and vision that have left a lasting imprint on several generations of scholars. For those who have worked closely with him, Budin’s academic work is inseparable from the intellectual curiosity, generosity and foresight that characterise his approach to research. With this Festschrift, we seek to honour both dimensions of his legacy: the conceptual contributions that continue to influence the international research community, and the personal, collegial and deeply human qualities that have touched so many of us. Our motivation to assemble this volume in honour of Gerhard Budin’s 65 th birthday and retirement, originates not only from admiration for his scholarly achievements but also from gratitude for his steadfast support and for the many ways he has shaped our own academic trajectories. It is a privilege to celebrate this milestone with him. As editors, we offer our warmest congratulations on his birthday and on the occasion of his well-earned retirement, and we hope that the contributions gathered here will serve both as a tribute to his past work and as an inspiration for future generations building on the foundations he has laid. <?page no="9"?> Worte der Wertschätzung Mira Kadrić Gerhard Budin hat das Zentrum für Translationswissenschaft der Universität Wien in den vergangenen Jahrzehnten in einzigartiger Weise geprägt. Sein Wirken vereint auf herausragende Weise das Engagement für Studierende, internationale Sichtbarkeit und ein hohes Verantwortungsbewusstsein in uni‐ versitären Leitungsfunktionen. Gerade diese Verbindung verleiht seinem akade‐ mischen Profil eine Wirkungskraft, die innerhalb wie außerhalb der Universität Wien höchste Anerkennung findet. Dazu kommt, dass Gerhard Budin für die Translationswissenschaft Zukunftsthemen - etwa Technikunterstützung und Digitalisierung - erschlossen hat, lange bevor alle davon gesprochen haben. Ein erstes zentrales Merkmal der Tätigkeit Gerhard Budins war die umfang‐ reiche Lehrtätigkeit sowie die Betreuung einer großen Zahl von Studierenden, insbesondere im Rahmen von Abschlussarbeiten. Über viele Jahre hinweg hat Gerhard Budin eine kaum zu überblickende Zahl an Masterarbeiten und Dissertationen begleitet - stets getragen von einer Haltung, die Orientierung gibt, zugleich aber Freiräume für eigene Ideen eröffnet und die Eigenverantwor‐ tung fördert. Gerade im Doktoratsstudium, wo über lange Zeiträume hinweg geforscht wird, hat Gerhard Budin als Mentor nie den individuellen Weg der Forschenden eingeschränkt. Damit hat er wesentlich dazu beigetragen, eine neue Generation von Wissenschafter: innen hervorzubringen, die fachlich exzellent und international konkurrenzfähig ist. Darüber hinaus hat Gerhard Budin im Laufe seiner wissenschaftlichen Lauf‐ bahn eine beeindruckende Zahl an Forschungsprojekten initiiert, geleitet und begleitet, die ein breites Spektrum von Themen abdecken - von der Termino‐ logiewissenschaft über die Digital Humanities bis hin zu sprachwissenschaftli‐ chen Großprojekten. Diese Projekte zeichnen sich durch Interdisziplinarität, internationale Vernetzung und nachhaltige Wirkung aus, indem sie nicht nur neue Erkenntnisse hervorbringen, sondern auch langfristige Infrastrukturen und offene Wissensressourcen schaffen. Durch seine Projekte und Engagement in zahlreichen internationalen Gremien, Verbänden und Netzwerken hat Ger‐ <?page no="10"?> hard Budin die Position der Universität Wien in der weltweiten Scientific Community nachhaltig gestärkt. Schließlich hat Gerhard Budin über viele Jahre hinweg in Leitungs- und Verwaltungsfunktionen Verantwortung übernommen - sei es als Institutsvor‐ stand, als Dekan oder in anderen universitären Gremien. Diese Tätigkeiten sind häufig wenig sichtbar, verlangen aber hohes organisatorisches Geschick, Ausdauer und Kompromissbereitschaft. Gerade in Leitungsfunktionen durfte ich selbst über viele Jahre hinweg eng mit Gerhard zusammenarbeiten. In dieser gemeinsamen Erfahrung zeigte sich, wie sehr Gerhard es verstand, strategische Entwicklungen auf den Weg zu bringen, die das Zentrum für Translationswissenschaft nachhaltig gestärkt haben. Die Verbindung dieser Dimensionen macht das Wirken Gerhard Budins so besonders. Er verkörpert ein akademisches Bild, das das Zentrum für Trans‐ lationswissenschaft und die Universität Wien entscheidend geprägt hat: die Förderung von Studierenden und Nachwuchswissenschaftler: innen, die wissen‐ schaftliche Führungsrolle in der internationalen Terminologiewissenschaft und die verantwortungsvolle Gestaltung universitärer Strukturen. Die Festschrift bietet einen wunderbaren Anlass, sein reiches Lebenswerk in seiner ganzen Breite sichtbar zu machen und zu würdigen. Es ist zugleich ein Moment, den Dank für die Energie und Weitsicht, mit der Gerhard Budin unser Fach, das Zentrum für Translationswissenschaft und die Universität Wien über viele Jahre geprägt hat, auszusprechen. Mit großer Wertschätzung und herzlichen Glückwünschen zum 65. Geburts‐ tag! Mira Kadrić 10 Mira Kadrić <?page no="11"?> 1 https: / / publikationen.radt.org/ RaDT_Terminologie_2015_deutsch.pdf 2 https: / / publikationen.radt.org/ RaDT_Wirtschaftspapier_2010_deutsch.pdf 3 https: / / publikationen.radt.org/ RaDT_Basiswissen_2013_deutsch.pdf 4 https: / / publikationen.radt.org/ RaDT_Terminologiewissenschaft_2017_deutsch.pdf Glückwunsch des Rats für Deutschsprachige Terminologie (RaDT) RaDT Gerhard Budin hat den Rat für Deutschsprachige Terminologie (RaDT) von Anfang an produktiv und engagiert begleitet: Er war in der Arbeitsgruppe für die deutschsprachige Terminologie-Initiative und somit an der Grundsteinlegung für den RaDT beteiligt, und er wirkte während der ersten zehn Jahre als dessen Schriftführer. Mit seinem feinen Gespür für Wissenstransfer und Kommunikation hat er bei der Erarbeitung der ersten RaDT-Broschüren mitgewirkt: So gab er wichtige Impulse für die erste Auflage des sog. Grundlagenpapiers («Termino‐ logie - Grundlage für Fachkommunikation, Fachinformation, Fachwissen»), das mittlerweile in dritter Auflage 1 verfügbar ist, für die Broschüre «Wissen, Marken, Kundenbindung - Kritischer Erfolgsfaktor Terminologie» 2 sowie für die Broschüre «Terminologisches Basiswissen für Fachleute» 3 . Unschätzbar war sein Beitrag zum RaDT-Positionspapier «Terminologiewissenschaft - Eine Standortbestimmung» 4 , das von seinem enormen Fachwissen profitiert hat. Als Gesprächspartner ist Gerhard unerreicht - er hat gute Einfälle, sieht Querverbindungen und bringt Einwände auf den Punkt, um danach gute Kom‐ promisslösungen zu entwickeln, die für alle stimmig sind. Der RaDT freut sich sehr, dass Gerhard Budin ihm weiter als RaDT-Freund verbunden bleibt, denn eine Terminologiewelt - vor allem eine deutschsprachige Terminologiewelt - ohne Gerhard Budin ist unvorstellbar. Lieber Gerhard, zum Eintritt in den Ruhestand wünschen wir Dir alles Gute! Deine aktuellen und ehemaligen Mitstreiterinnen und Mitstreiter im Rat für Deutschsprachige Terminologie <?page no="13"?> 1 Galinski, Christian (2019). Blütezeit der Zusammenarbeit zwischen Terminologie ei‐ nerseits und Information und Dokumentation andererseits 1980-2000. In: Drewer. Petra/ Pulitano, Donatella (Hrsg.). Terminologie: Epochen - Schwerpunkte - Umset‐ zungen. Zum 25-jährigen Bestehen des Rats für Deutschsprachige Terminologie. Berlin, 21-43. 2 Bühler, Hildegund (1998). The scientific legacy of Eugen Wüster revisited: Three major unfinished projects. In: Oeser. E./ Galinski, C. (Hrsg.). Eugen Wüster (1898-1977). Leben und Werk. Ein österreichischer Pionier der Informationsgesellschaft / His life and work. An Austrian pioneer of the information society. Wien, Infoterm, 1998. 187-205. Erinnerungen an 15 Jahre intensiver Zusammenarbeit mit Prof. Dr. Gerhard Budin 1987~2001 Christian Galinski 2026 lädt zu Rückblicken ein: Gerhard Budin wurde vor 65 Jahren geboren, Infoterm wurde knapp zehn Jahre später gegründet. Vor 40 Jahren trafen sich die Zeitlinien, und es begann eine Phase unvergesslich intensiver Zusam‐ menarbeit mit Gerhard. Diese fiel auch in die „Blütezeit der Zusammenarbeit zwischen Terminologie und IuD“ 1 , eine Zeit wesentlicher wissenschaftlicher Entwicklungen und Aktivitäten, auch im Bereich der technischen Normung. Sie war geprägt durch die Fortführung der Aktivitäten Eugen Wüsters auf vielen Gebieten sowie durch die Rettung und Aufstellung seines Nachlasses. Letzteres begann bereits zwei Jahre nach Wüsters Ableben (1977) 2 und wurde vor allem ab 1985 durch Infoterm-Mitarbeiter und Kooperationspartner intensiv weitergeführt. Dazu musste der umfangreiche Nachlass Wüsters in Wieselburg (Niederösterreich) zunächst vor der Entsorgung gerettet und anschließend für die weitere Bearbeitung bei Infoterm geordnet aufgestellt werden. Dies geschah im Wesentlichen in den 1980er und 1990er Jahren, wobei Gerhard sich an manchen dieser Aktionen bereits in der zweiten Hälfte der 1980er Jahre aktiv beteiligte. <?page no="14"?> 3 Später Austrian Standards Institute, heute bekannt als Austrian Standards International (ASI). 1971~1996 war Infoterm eine Abteilung des ON. 4 Nedobity, Wolfang (1998). Der wissenschaftliche Briefwechsel Eugen Wüsters. In: Oeser; E./ Galinski, C. (Hrsg.). Eugen Wüster (1898-1977). Leben und Werk. Ein österrei‐ chischer Pionier der Informationsgesellschaft / His life and work. An Austrian pioneer of the information society. Wien, Infoterm, 175-185. Die Bekanntschaft mit Gerhard ergab sich um 1985, vielleicht über die von Infoterm initiierten Gemeinschaftsseminare oder während einer der seit 1980 jährlich stattfindenden Terminologiesommerschulen (TSS). Gerhard, der gerade seinen Magistertitel erworben hatte, gefiel der integrative Ansatz in Terminologielehre und terminologischen Aktivitäten sowie zu benachbarten Disziplinen, den Infoterm verfolgte. In den Gemeinschaftsseminaren trafen sich österreichische (später auch deutsche) Universitätslehrer und ihre Studierenden der Fachbereiche Wissenschaftstheorie, Übersetzen und Dolmetschen, Kommu‐ nikationswissenschaft, Informationswesen und Computerwissenschaft, um das Thema „Terminologien in der Fachkommunikation“ integrativ zu diskutieren. Gerhard stieß zu Infoterm in einer Zeit, als sich Infoterm der Vielzahl von Besuchern aus aller Welt, Einladungen zu Veranstaltungen in alle Welt und Bewerbungen um eine Stelle oder ein Volontariat kaum erwehren konnte. Die Rettung des Wüster-Nachlasses erfolgte schrittweise. Zuerst wurden seine Literatursammlung und das umfangreiche Archiv mit etwa 100 lfm Ord‐ nern voller Dokumente nach Wien transportiert und im Keller des Österreichi‐ schen Normungsinstituts (ON 3 ) eingelagert. Einige Monate später übersiedelte Infoterm in eine ON-Nebenstelle, wo der bisher übernommene Teil des Nach‐ lasses aufgestellt werden konnte. Schon ein oder zwei Jahre später beteiligte sich Gerhard am Transfer des 90 Ordner umfassenden Briefwechsels Wüsters 4 sowie der Karteien seiner Terminologie- und Lexikographieprojekte von Wie‐ selburg nach Wien zu Infoterm. Dabei konnten auch Daten und Dokumente zu seinen Esperanto- und anderen Plansprachen-Aktivitäten gesichert und dem Esperanto-Museum der Österreichischen Nationalbibliothek übergeben werden. Bald nach seiner Teilnahme an der TSS und den Gemeinschaftsseminaren um 1985 begann Gerhard, sich in verschiedenen Infoterm-Aktivitäten zu enga‐ gieren, darunter die Mitarbeit in Projekten mit dem Europäischen Zentrum für berufliche Weiterbildung (CEDEFOP), die Mithilfe bei der Organisation des ersten internationalen Kongresses über „Terminologie und Wissenstechnik“ (TKE ’87), die Mitarbeit in Arbeitsgruppen der ursprünglich zum Zweck der Abhaltung der TKE-Kongresse gegründeten Gesellschaft für Terminologie und 14 Christian Galinski <?page no="15"?> 5 Felber, Helmut/ Budin, Gerhard (1989) Terminologie in Theorie und Praxis. Tübingen: Gunter Narr Verlag. 6 Felber, Helmut (1990). Allgemeine Terminologielehre und Wissenstechnik. Theoreti‐ sche Grundlagen. 2. Auflage, Wien: TermNet. Wissenstransfer (GTW) sowie die Gründung 1989 des Internationalen Instituts für Terminologieforschung (IITF), dessen erster Generalsekretär er wurde. Im Rahmen des IITF gehörte Gerhard auch zu den Mitinitiatoren des wis‐ senschaftlichen Periodikums Terminology Science & Research, das jahrelang substanzielle Beiträge brachte. Das 1988 als Verein gegründete Internationale Terminologienetz (TermNet) wurde 1992 mit einer Anschubfinanzierung aus Infoterm-Projektgeldern selbständig, übersiedelte an einen neuen Standort und übernahm fortan die Produktion und Publikation der Infoterm- und TermNet-Periodika sowie der IITF-Publikationen. So entstand - einschließlich ISO/ TC 37 - eine informelle „Vierergruppe“, die Aufgaben und Aktivitäten auf internationaler Ebene schwerpunktmäßig aufteilte und koordinierte. Wegen ihrer Treffen im Café Hummel wurde sie von uns auch „Hummelgruppe“ genannt. Bewundernswerterweise fand Gerhard trotz seines Engagements die Zeit für die Zusammenarbeit mit meinem Vorgänger Helmut Felber beim Verfassen der gemeinsamen Buchpublikation Terminologie in Theorie und Praxis  5 sowie zur Beratung Felbers bei der ersten Auflage seines Buches Allgemeine Terminologie‐ lehre und Wissenstechnik  6 gegen Ende der 1980er Jahre. Unglaublich, wie er zu alledem im Juni 1988 auch noch das Doktorat zum Abschluss seines Studiums schaffte. 1990 konnte Gerhard als offizieller Mitarbeiter von Infoterm gewonnen werden, wodurch die Zusammenarbeit schlagartig an Intensität zunahm. Von Anfang an engagierte er sich mit zunehmender Energie in allen Aktivitäten und Projekten - von Normungsaktivitäten auf nationaler und internationaler Ebene über Publikationsprojekte, Zusammenarbeit mit Stellen im In- und Ausland, Organisation von Veranstaltungen und Ausbildungen bis hin zu verschiedenen Forschungs- und Entwicklungsprojekten sowie Beratungsaktivitäten. Schon bald fungierte er de facto, ab 1994 offiziell, als mein Stellvertreter. Seine Aktivitäten erweiterten sich vor allem um Normungsaktivitäten in den Be‐ reichen Terminologie, Information und Dokumentation sowie Klassifikation. Darüber hinaus initiierte und koordinierte er wissenschaftliche Projekte und Ausbildungsaktivitäten - neben seinen Lehrverpflichtungen an der Universität Wien. Nicht zuletzt dank seiner Anstellung konnte Infoterm seit 1990 umfassend all seine Hauptfunktionen und -tätigkeiten wahrnehmen. Diese umfassten alle Erinnerungen an 15 Jahre intensiver Zusammenarbeit mit Prof. Dr. Gerhard Budin 15 <?page no="16"?> 7 TermNet. (Hrsg.). 9-sprachiges Thesaurus-Erstellungs- und Verwaltungs-Programm unter besonderer Berücksichtigung der osteuropäischen Sprachen. Forschungsauftrag des Bundesministeriums für Wissenschaft und Forschung. Endbericht. Wien: 1991 (Infoterm Dokument 17-1995). 8 CEDEFOP (Hrsg.). Multilingual Thesaurus of vocational training. Berlin, 1992. 9 OECD (Hrsg.). Macrothesaurus for Information Processing in the Field of Economic and Social Development, 1991. Facetten der Terminologiedokumentation, Terminologienormung und Förde‐ rung der internationalen Zusammenarbeit sowie damit zusammenhängende wissenschaftliche oder FuE-Projekte, Veranstaltungen, Beratungs- und Ausbil‐ dungsaktivitäten. Einige der erfolgreichen Resultate dieser Zeit, zu denen Gerhard maßgeblich beitrug, waren: - Weiterentwicklung der Methodik der Terminologiedokumentation und Herausgabe mehrerer damals dringend benötigter Bibliografien und Hand‐ bücher, entsprechende Ausbildungs- und Beratungsaktivitäten sowie Soft‐ wareentwicklungen, - Führung des ISO/ TC 37-Sekretariats und Initiierung sowie Koordination vieler Normungsaktivitäten, einschließlich der gründlichen Überarbeitung existierender Normen, - Förderung der Zusammenarbeit auf dem Gebiet der Terminologie und benachbarter Gebiete in Ländern Afrikas, Amerikas, Asiens, Europas und der arabischen Welt. Die Vielzahl dieser Aktivitäten und ihrer Erfolge ist durch Infoterm Dokumente - vor allem die Infoterm Annual Reports - gut dokumentiert. Vieles davon ist über reguläre Publikationen in Form von Artikeln und Büchern zugänglich. Die Zusammenarbeit mit TermNet und IITF erwies sich als äußerst zielführend für alle Beteiligten. Gerhard setzte sich unermüdlich für den Erfolg dieser Aktivitäten ein, was auch mit viel Reisen verbunden war. Oft trafen wir uns - aus verschiedenen Gegenden kommend - zu gemeinsamen Aktivitäten in Beijing, Ottawa, Brüssel, Paris, Kopenhagen oder Luxemburg, um nur einige Orte zu nennen. Dabei ging es häufig um neue formale und informelle Infrastrukturen, wodurch auch Gerhard wesentlich zum Entstehen von Terminologie-Netzwer‐ ken, Institutionen und Organisationen in verschiedenen Teilen der Welt beitrug. Infoterm engagierte sich über Verträge in Software-Entwicklung auf Grund‐ lage des UNESCO-Bibliotheksprogramms CDSISIS (Version 3 MicroISIS) 7 und nutzte die Resultate für die Revision des CEDEFOP-Thesaurus 8 und des OECD-Macrothesaurus 9 . Gemeinsam mit Gerhard entwickelten wir integrative Konzepte wie T&D (Terminologie und Dokumentation), die terminologische 16 Christian Galinski <?page no="17"?> 10 ISO/ IEC Directives, Part 2. Principles and rules for the structure and drafting of ISO and IEC documents. 11 Study on the re-assessment of the original mandate of Infoterm.” In: Infoterm Annual Report 1993, p. 4. Taschen-Wissensdatenbank (für das japanische Softwarehaus INS Corp.) und OpenMLCM (Open Multilingual Content Management) für die UNESCO. Diese Entwicklungen hatten Einfluss auf die Normungsarbeit in ISO/ TC 37 - umge‐ kehrt wurde größter Wert auf Normkonformität zu internationalen Normen gelegt. Dies zeitigte einigen Einfluss auf andere Komitees in ISO und IEC bis hin zur Weiterentwicklung des ISO-Portals. Detaillierte Regeln für die Terminolo‐ gienormung fanden u. a. dank der ersten Ausgabe der Norm ISO 10241-1: 1992 International Terminology Standards - Preparation and layout Eingang in Teil 2 der ISO/ IEC-Richtlinien für die Normungsarbeit 10 . Sowohl daran als auch an der Nutzung der oben genannten Projektresultate für die erfolgreiche Entwicklung einer ON-Terminologiedatenbank war Gerhard wesentlich beteiligt. Trotz der Gründung von TermNet und IITF und der dadurch ermöglichten Aufgabenteilung wurde Infoterm von seinem Erfolg regelrecht überrollt. In manchen Monaten kamen fast täglich Besucher. Der Tausch von Publikationen und Fachzeitschriften füllte Regale. Ständig waren ein bis drei Freiwillige als Volontär: innen mit verschiedenen Aufgaben beschäftigt. Auch kamen Infoterm in Erfüllung seines ursprünglichen UNESCO-Mandats und angesichts der all‐ gemeinen wissenschaftlich-technologischen Entwicklungen - insbesondere im Bereich der Informations- und Kommunikationstechnologien (IKT) - im Laufe der Jahre immer mehr Aufgaben und Rollen zu. Zunehmend wurde erkannt, dass Infoterm ohne eine gründliche Restrukturierung an unüberwindliche Grenzen stoßen würde. 1994 verfasste ich gemeinsam mit Gerhard im Auftrag des österreichischen Wirtschaftsministeriums die Studie zur künftigen Rolle Infoterms für die euro‐ päische Terminologie-Infrastruktur. Bereits 1992 hatte die UNESCO eine inter‐ nationale Expertengruppe mit einer Studie zur Neubewertung des bisherigen (1971-1992) Mandats von Infoterm  11 beauftragt. Gerhard versorgte die Gruppe mit Unterlagen und fachlichem Rat, sodass die positiven Resultate 1993 in der Sitzung des Infoterm-Beirats (IAC) diskutiert und anschließend mit Empfehlun‐ gen des IAC an die UNESCO und die zuständigen österreichischen Stellen übermittelt werden konnten. Empfohlen wurde u. a. ein eigener rechtlicher Status für Infoterm - ein Gedanke, der bereits in den 1960er und 1970er Jahren im Raum gestanden hatte. Alles schien sich in eine wünschenswerte Richtung zu entwickeln. Der für Infoterm zuständige stellvertretende UNESCO-Generaldirektor besuchte Erinnerungen an 15 Jahre intensiver Zusammenarbeit mit Prof. Dr. Gerhard Budin 17 <?page no="18"?> Infoterm 1993, um sich ein Bild zu machen. Im Vorfeld des erwarteten EU-Bei‐ tritts Österreichs begannen sich auch EU-Kommissare für „Terminologie“ zu interessieren - insbesondere im Zusammenhang mit der Weiterentwicklung automatischer Übersetzungssysteme. Schon vor dem offiziellen EU-Beitritt am 1. Januar 1995 war Infoterm in verschiedenen Beratungsgremien der EU-Kom‐ mission vertreten, etwa bei der Vorbereitung des Vierten Rahmenprogramms der Europäischen Gemeinschaften im Bereich Forschung, technologische Ent‐ wicklung und Demonstration (1994-1998). Gerhard vertrat Infoterm in EU-Projekten wie Translator’s Workbench (1989- 1992) oder POINTER: Proposals for an Operational Infrastructure for Terminology in Europe (1994-1995). Kooperationen mit ausländischen Organisationen und Institutionen wurden weiterentwickelt oder neu angebahnt. Die Ausbildungs- und Beratungsaktivitäten nahmen - nicht zuletzt dank Gerhards Engagement - deutlich zu. In der zweiten Jahreshälfte 1995 trübten sich die guten Aussichten, und Ende 1995 schien Infoterm vor dem Aus zu stehen. Anfang 1996 wurde Personal abgezogen oder entlassen; Gerhard verließ Infoterm, um an der Universität Wien seine akademische Karriere fortzusetzen. Dank größter Anstrengungen des ON und anderer österreichischer Stellen konnte jedoch in letzter Minute eine Finanzierung in Österreich gesichert werden, die auch eine Anschubfinan‐ zierung für die Neueinrichtung eines Büros umfasste. So konnte Infoterm am 1. Juli 1996 Räume in einem Innovationszentrum im Süden Wiens beziehen. Im August 1996 erfolgte die Gründung Infoterms als selbständiger interna‐ tionaler wissenschaftlicher Verein nach österreichischem Recht. Prof. Dr. Erhard Oeser von der Universität Wien wurde erster Präsident von Infoterm. Prof. Oeser, der auch formell Gerhards akademischer Vorgesetzter war, zeigte großes Interesse an den Aktivitäten Infoterms. Es dauerte etwa ein Jahr, bis die Räumlichkeiten am neuen Standort so einge‐ richtet waren, dass das Wüster-Archiv, das Normenarchiv, die Büchersammlung und der Dokumentationsbereich optimal geordnet und für Mitarbeiter und Be‐ sucher nutzbar waren. Gleichzeitig mussten Aktivitäten wie die Betreuung der Normungsagenden im Auftrag des ON oder lang vorgeplante Veranstaltungen durchgeführt werden, auch wenn Unterlagen wegen der noch fehlenden Schie‐ beregale zunächst nur teilweise aufgestellt werden konnten. Trotz widriger Umstände konnte die 4. TKE ’96 - nicht zuletzt dank Gerhards tatkräftiger Unterstützung - erfolgreich in Wien ausgerichtet werden. Neue Mitarbeiter mussten angestellt und eingeschult werden. Unglaublich, wie Gerhard es in dieser Zeit - sicherlich nicht „nebenbei“ - schaffte, sich zu habilitieren und eine Professur zu übernehmen. 18 Christian Galinski <?page no="19"?> 12 Hoffmann, Lothar/ Kalverkämper, Hartwig/ Wiegand, Herbert Ernst (in Verbindung mit C. Galinski und W. Hüllen) (Hrsg.). (1998). Fachsprachen / Languages for Special Purposes. Ein internationales Handbuch zur Fachsprachenforschung und Terminolo‐ giewissenschaft / An International Handbook of Special Language and Terminology Research. Berlin/ New York: De Gruyter, (HSK 14.2). 13 The European Network of Terminology Information and Documentation Centres (TDCnet) 14 Wright, Sue Ellen/ Budin, Gerhard (eds.) (1997). Handbook of terminology manage‐ ment. Vol. 1: Basic aspects of terminology management. Amsterdam/ Philadelphia: John Benjamins. Ende 1997 konnte begonnen werden, die Infoterm- und TermNet-Zeitschrif‐ ten sowie neue Buchpublikationen aus Kostengründen wieder bei Infoterm zu produzieren und verschiedene Aktivitäten ungehindert fortzuführen bzw. neu anzubahnen. Gerhard hatte TermNet bereits zuvor bei Publikationen unterstützt und setzte diese Unterstützung auch am neuen TermNet-Standort bei Infoterm fort. Für die Jahre 1996 bis 2002 möchte ich drei Aktivitäten besonders hervor‐ heben, die von intensiver und überaus erfolgreicher Zusammenarbeit mit Gerhard geprägt waren und für die Entwicklung der Terminologiewissenschaft und terminologischer Aktivitäten weltweit große Bedeutung hatten: 1. die Organisation der Beiträge zur Buchpublikation Fachsprachen. Ein internationales Handbuch zur Fachsprachenforschung und Terminologiewissenschaft  12 , 2. die Auf‐ stellung und Bearbeitung des Wüster-Archivs, und 3. das EU-Projekt TDCnet  13 . 1995 kontaktierte Prof. Wiegand aus Deutschland Infoterm mit der Anregung, ein Kapitel mit Beiträgen zur Terminologiewissenschaft und Terminographie zu organisieren. Da ich angesichts der prekären Umstände keine Zusage machen konnte, kam Prof. Wiegand im zweiten Halbjahr 1996 nach Wien, um Details zu besprechen. Als ich Gerhard davon erzählte, sagte er spontan: „Das muss Infoterm machen - ich helfe euch! “ Auf Basis eines Themenvorschlags wurden unter großem Zeitdruck zahl‐ reiche potenzielle Autor: innen kontaktiert. Entstanden sind Kapitel XXIII (Die Terminographie im 20. Jahrhundert), Kapitel XXIV (Terminologiearbeit und Terminologieregelung) sowie mehrere Beiträge in anderen Kapiteln über Fachsprachen und Fachkommunikation, jeweils mit angemessener Berück‐ sichtigung terminologischer Aspekte. Parallel dazu arbeitete Gerhard gemein‐ sam mit S. E. Wright am Handbook of Terminology Management  14 . Die Ver‐ öffentlichungsstrategie Infoterms hatte bereits zuvor andere Publikationen stimuliert, etwa die Gründung der Zeitschrift Terminology beim Verlag John Benjamins. Die Beiträge über Terminologie in den genannten Handbüchern Erinnerungen an 15 Jahre intensiver Zusammenarbeit mit Prof. Dr. Gerhard Budin 19 <?page no="20"?> 15 Infoterm. Report of the 5th Session of the Infoterm General Assembly. 2001. 16 Multilingual Information Society (MLIS) - Multiannual programme to promote the linguistic diversity of Europe in the Information Society, 1996-1998. 17 Kifer, E. (1999). Klassifikationsbereinigung als Voraussetzung für terminologische Mehrwertprodukte. 18 Budin, Gerhard/ Körbisch, P./ Szpott, H. (2000). Terminologieprodukte: Produkt- und Marktübersicht. waren jedenfalls - dank Gerhards ausschlaggebender Mitwirkung - in hohem Maße bahnbrechend für das Fachgebiet. Nach der Einrichtung eines eigenen Raums für das Wüster-Archiv konnten die thematisch beschrifteten Ordner in der originalen Wüster’schen Ordnung aufgestellt werden. Gerhard arbeitete dabei intensiv mit meinem Vorgänger Helmut Felber zusammen, der sich mit großer Freude dieser Aufgabe widmete. Über kleinere Projekte wurden einige von Wüster angelegte Karteien - etwa sein Key to Terminology - mikroverfilmt und mit Gerhard besprochen, wie das Archiv insgesamt am besten für die Nachwelt erhalten werden könnte. 1999 wurde das Archiv samt Regalen und Mobiliar dem Institut für Wissen‐ schaftstheorie und Wissenschaftsforschung der Universität Wien übergeben. Prof. Oeser, Leiter dieses Instituts und Infoterm-Präsident, konnte 2001 in der Infoterm-Generalversammlung stolz verkünden, dass „das Eugen Wüster-Ar‐ chiv zusammen mit dem Sir Karl Popper-Nachlass nun Teil des Instituts-For‐ schungsarchivs ist und erstmals Studierenden und der Öffentlichkeit zugänglich gemacht wurde“. Im gleichen Atemzug dankte er „Gerhard und seinen Mitar‐ beitern für die ausgezeichnete Betreuung des Archivs“ 15 . Die dritte Aktivität betrifft das EU-Projekt European Terminology Documenta‐ tion Centre Network (TDCnet) im Rahmen des EU-Programms MLIS 16 . Dank die‐ ses Projekts (1998-2000) konnten alle bibliographischen Daten der Bibliotheks- und Dokumentationsbestände Infoterms erfasst und in einem einheitlichen Datenformat dem Projektkonsortium übergeben werden. Gerhard unterstützte viele der im Projekt anfallenden Aufgaben. So betreute er u. a. Untersuchungen durch Studierende zu einzelnen Projektthemen, etwa zur Entwicklung einer einheitlichen Dachklassifikation für Terminologiezwecke 17 oder zu einer Studie über die Aussichten auf dem Terminologiemarkt 18 . 2001 wurden die Bibliotheksbestände Infoterms dank der Unterstützung durch Prof. Klaus-Dirk Schmitz zur neuen Infoterm Zweigstelle an der Fach‐ hochschule Köln übersiedelt, wo sie besser genutzt und weiteren Kreisen zugänglich gemacht werden konnten. Ab September 2002 konnte sich Infoterm wieder auf Normungsaktivitäten im Auftrag des ON, Beratungstätigkeiten und 20 Christian Galinski <?page no="21"?> FuE-Projekte zur Weiterentwicklung der Terminologiewissenschaft und ihrer praktischen Anwendungen konzentrieren. Die Zusammenarbeit mit Gerhard setzte sich auch nach der Jahrtausend‐ wende fort - vor allem im Bereich der Terminologienormung -, obwohl sich Gerhards akademische Karriere und Infoterms Aktivitätsschwerpunkte zuneh‐ mend in verschiedene Richtungen entwickelten. Mal trafen wir uns in Projekten wieder, mal besprachen wir Beiträge für Publikationen oder Veranstaltungen - der Austausch riss jedenfalls nie ab. Die obigen Erinnerungen können nur einen Bruchteil der teils dramatischen, aber stets sehr dynamischen Entwicklung Infoterms in den 15 Jahren intensivs‐ ter Zusammenarbeit (1987-2001) mit Gerhard wiedergeben. 2018 wurde ihm auf Beschluss des Infoterm Vorstands der Eugen Wüster-Preis für seine vielfältigen Verdienste verliehen. Für die hier im Zeitraffer wiedergegebene Zusammenarbeit über all die Jahre möchte ich ihm persönlich und auch im Namen von Infoterm danken und ihm zu seinem 65. Geburtstag von Herzen alles Gute wünschen. Christian Galinski Erinnerungen an 15 Jahre intensiver Zusammenarbeit mit Prof. Dr. Gerhard Budin 21 <?page no="23"?> Beyond Terminology: A tribute to Gerhard Budin Dagmar Gromann, Barbara Heinisch & Vesna Lušicky The title of this volume is a tribute to as well as a characteristic of the scholarly life it celebrates. Beyond Terminology captures the notion that the work of Gerhard Budin, even though centrally contributing to the field of terminology science, has never been confined to the boundaries of a single discipline. This interdisciplinary nature of his academic career already surfaced early on in his PhD thesis Möglichkeiten und Grenzen der Terminologieforschung in den Sozialwissenschaften (1988) that he completed in linguistics at the University of Vienna, which applies terminological and linguistic analyses to social sci‐ ences and later cumulated in a book (Budin 1993a). Budin himself highlighted the highly interdisciplinary nature that terminology science requires in his habilitation that lead to his seminal work and book Wissensorganisation und Ter‐ minologie (Budin 1996a) on the integration of information science, knowledge representation and organisation, and terminology in the form of his famous Wissens-, Informations- und Kommunikations-Organisation (Knowledge, In‐ formation and Communication Organisation; WIKO) model. The international influence of this idea is evidenced in this volume, e.g. by Juan Carlos Diaz Vásquez and Maria Cecilia Plested Alvarez. Terminology in all its facets has certainly been a lifelong thread running through his research. Yet anyone who has followed his career across four decades will know that this thread has always been interwoven in a much larger academic fabric encompassing the philosophy of language, language technologies, machine translation, cognitive computing, corpus linguistics, knowledge engineering, digital humanities, and the governance of international standards. This represents a quite extensive albeit not complete list of interdisciplinary endeavours that he has undertaken throughout his research career. To honour Gerhard Budin is therefore not only to honour a pioneer in the field of terminology, but a scholar who has consistently ensured a sustained and constitutive dialog with adjacent discipline. <?page no="24"?> The contributions assembled in this volume reflect on this interdisciplinary academic trajectory as well as its influence. They are the original work of colleagues, collaborators, and former students from across the international community of terminology science, translation studies, language technology, digital humanities, and transdisciplinary linguistics. Gerhard Budin has been mentoring and assisting students and young researchers for decades. All three of us were lucky to be supervised by Gerhard Budin and inspired by him to pursue a research career, as his mode and manner of work at the Centre for Translation Studies at the University of Vienna and in national as well as international collaborations has always been truly a source of inspiration. We were only three among the astonishingly large number of fortunate ones to have him as a PhD thesis supervisor, let alone the extraordinarily large number of master’s theses he supervised, which are estimated as more than 400. We take the occasion of compiling this volume to celebrate Gerhard’s 65 th birthday and retirement and to express our deepest gratitude for being a truly inspiring mentor and researcher to so many, apart from being a very kind, warm, and congenial person. We will continue this introduction with a brief account of Gerhard Budin’s astonishing academic life and work, as a complete account would require a volume on its own. Gerhard Budin was born in Vienna in 1961, which was equally the formative city of his career and research. His intellectual formation is indeed inseparable from the city and environment in which it took place. Vienna is internationally recognized as a centre of philosophical and terminological thinking and the place at which Eugen Wüster (1967, 1971) developed the foundational framework for terminology science that would become to be known as the General Theory of Terminology (Wüster 1974). And it was here that the institutional infrastructure of the discipline took shape with the International Information Centre for Terminology (Infoterm), to which Gerhard Budin actively contributed over many years as described in detail by Christian Galinski in this volume. In the early 1980s he was a student of translation studies, linguistics, and the philosophy of science in Vienna and Barcelona, Spain. Since 1990, he worked as research assistant at Infoterm as well as lecturer in terminology science and translation-oriented terminology work at the University of Vienna, after a research assistant position at the Institute of the Philosophy of Science at the same university in 1988. Even before officially joining Infoterm as an employee in 1990, he already published with Helmut Felber on terminology in theory and practice (Felber and Budin 1989), a collaborator of Eugen Wüster, director of Infoterm, and secretary of International Standards Organisation, Technical Committee 37 (ISO/ TC 37) as evidenced by Christian Galinski in this volume. Simultaneously, Gerhard 24 Dagmar Gromann, Barbara Heinisch & Vesna Lušicky <?page no="25"?> Budin became the first Secretary-General of the newly founded International Institute for Terminology Research (IITF), a forum for exchanging research and teaching experiences in the field of terminology. At that time Infoterm was not only housed at the Austrian Standards Institute, but also had strong ties to standardisation. These were shared by Gerhard Budin, as he had already been a general member and national delegate to the ISO/ TC 37 on Terminology and other Language and Content Resources since 1985, assuming several long-lasting chair positions from 2001 onwards. His research activities were, however, by no means confined by national boundaries. Gerhard Budin not only actively taught repeatedly in Terminology Summer Schools, but also helped to organise international conferences, inclu‐ ding the first Terminology and Knowledge Engineering (TKE) conference in 1987. Apart from contributing actively to a later edition (Budin 1996b), he also directly supported the organisation of TKE in Vienna in 1996. He also organised conferences on Language for Specific/ Special Purposes (LSP), such as on Multilingualism and Specialist Communication (Budin 1995) or LSP (Budin and Lušicky 2014; Lušicky and Budin 2016), the latter of which twice jointly with a co-author of this introduction. His early conference contributions continued to extol the breadth of his scientific palette, e.g. Lukas et al. (1993) on how hypermedia systems could represent culturally specific knowledge from multiple perspectives simultaneously. In the same year and volume, Budin (1993b) published early work on integrating knowledge organisation and terminological modelling. Budin’s engagement with the relationship between translation studies and terminology was articulated in several significant articles. Galinski and Budin (1993) examine the practical challenges and needs of terminology management particularly focused on translation. Budin (1994) offered an early formulation of the argument that translation studies and terminology science share a set of fundamental concerns about specialised knowledge, cross-linguistic equiva‐ lence, and expert communication that neither discipline can adequately address in isolation. Knowledge about terminology can be considered a key competence of translation (Budin 2016b, 2016c). A complementary contribution focused on the structure and need for standardisation of terminological entries was jointly published with Sue Ellen Wright (Wright and Budin 1994), already showing a strong interest in data interoperability and accessibility. The theoretical culmination of this first phase of Gerhard Budin’s career is the monograph Wissensorganisation und Terminologie (1996) that draws on the philosophy of science, and epistemology alongside terminology theory to develop an interdisciplinary model for the organisation of knowledge, Beyond Terminology: A tribute to Gerhard Budin 25 <?page no="26"?> information and communication called WIKO. What is striking about the work is not only its disciplinary range but its explicit engagement with the epistemological dimensions of knowledge organisation. It seeks to provide a stronger theoretical basis for what Wüster called a unit of knowledge and to investigate the nature and structure of specialised knowledge. Gerhard Budin would later on continuously revisit this major topic from different perspectives, especially from a philosophical viewpoint, e.g. in comparing Brentano’s object theory with the conceptual framework of Wüster (Budin 2007). The topic of how to structure, align, communicate, and share data, information, and knowledge would accompany his research for the following decades. In the following year, Budin jointly with Erhard Oeser brought together contributions on the interface between terminology science and knowledge engineering (Budin and Oeser 1997), and that reflects the sustained intellectual collaboration between Budin and the philosophical tradition represented by Oeser’s work on the history and epistemology of science. Budin co-edited the Festschrift for Oeser together with Stephan Haltmayer and Franz Wuketits (Budin et al. 2000). In his acclaimed contribution Wüster, Schmitz und die Folgen (2018) to Klaus-Dirk Schmitz’s Festschrift, who also contributes to this volume, Gerhard Budin skillfully presents a detailed account of the development of terminology science from Wüster to Schmitz and the theoretical basis of terminology science as modern, multidimensional knowledge management. In Budin (2019) he argues that this interdisciplinary and multiperspective character of terminology science strengthens the discipline. This period also saw the completion of two major collaborative volumes that, to this date, remain his most widely cited publications. The two-volume Hand‐ book of Terminology Science (Wright and Budin 1997, 2001) brings together leading experts in the field of terminology management from theory to practice. Both editors contributed to chapters in both volumes. Aside from an existing translation of the volumes to Chinese, it still represents the standard work of reference for research and teaching in terminology management internationally. A second major line of work from this period relates to the management and accessibility of language resources and technology. One of the major challenges of language resources at the time was a common way to access them, which is why Budin coordinated the EU-funded project Standards-based Access to Multilingual Lexical and Terminological Resources (SALT) from 2000 to 2003. To increase the accessibility and compatibility of language resources, the SALT project contributed to developing standardised interchange formats (Budin and Melby 2000). Budin (2001, 2010a) depicts computational processing of terminological data as a paradigm shift for the discipline that requires a 26 Dagmar Gromann, Barbara Heinisch & Vesna Lušicky <?page no="27"?> reconsideration of theoretical frameworks. A parallel initiative that also raised similar concerns was the community of the Semantic Web, which Budin (2006) also considered for the interchange and accessibility of terminological data. Both initiatives have provided standardised interchange formats for language data, where the work of Budin, in particular within ISO/ TC 37, has had a stronger focus on terminology-specific requirements. After an associate professorship and short-term visiting professorships at the University of Limoges and Paris Diderot University, Gerhard Budin was appointed Full Professor for Translation-oriented Terminology Studies and Translation Technologies at the Centre for Translation Studies of the University of Vienna in 2005, of which he immediately became Vice Dean, followed by a Dean position from 2006 to 2008. Nevertheless, he continued his activities as visiting professor to foster international exchange, e.g. with the City University of Hong Kong in 2012. In 2010, he founded the Institute for Corpus Linguistics and Text Technology (ICLTT) at the Austrian Academy of Sciences and was its director until 2013, which marked the institutionalisation of his engagement in digital research infrastructures and digital humanities. It was only natural to later integrate the institute with what is now called the Austrian Centre for Digital Humanities represented in this volume by Tanja Wissik. His interest in providing a standar‐ dised access method to data and technologies was also evident in a Language Resources and Evaluation Conference (LREC) workshop he jointly organised in 2010 (Budin et al. 2010b). It is quite astonishing that in the same time period from 2010 to 2013, Gerhard Budin still had the stamina to coordinate the Austrian project contributing to the Common Language Resources and Technology Infrastructure (CLARIN) initiative of the European Research Infrastructure Consortium (ERIC) financed by the then Federal Ministry of Science and Research of Austria. This project and his ICLTT activities were well aligned in the sense of their relation to developing digital infrastructures. This interest and quest for interoperable language resources also led Budin to coordinate the EU-funded project Legal Language Interoperability Services (LISE) from 2011 to 2013, which addressed the urgent need to coordinate legal and administrative terminologies across the languages of European institutions, establishing interoperability among multi‐ lingual terminological databases through web-based terminology management workflows. His sustained concern with the technological and policy dimensions of multilingual communication goes well beyond technological optimisation (Rehm et al. 2020; Lušicky and Budin 2024). Beyond Terminology: A tribute to Gerhard Budin 27 <?page no="28"?> His outstanding track record of simultaneously juggling leadership positions, collaborations, publications, and managing projects continued throughout his career, not to mention the extensive supervision and teaching activities. For instance, recently, he was coordinator of the FWF-funded project Exil: Trans (2019-2023) which investigates and tracks the life and work of translators in exile. He also led the subproject Collaborative Online Research Platform within the larger FWF-funded project German in Austria. Variation - Contact - Percep‐ tion (2016-2026) as depicted in Budin et al. (2019), among many other projects and initiatives. This shows that his research has truly remained interdisciplinary throughout his career. The chapters collected in Beyond Terminology reflect the breadth of his scholarly influence and the diversity of the fields he helped shape. From diffe‐ rent perspectives, they demonstrate how his work on terminology, language technology, multilingual communication and translation continues to inform and inspire current research. In the first section that is devoted to terminology science, several contributions revisit central questions that have accompanied Budin’s work throughout his career. -Klaus Dirk Schmitz highlights the development of standards and structured terminological data, with particular attention to concept systems and their representation in terminological data‐ bases and TBX. Rute Costa discusses the difference between terminological and specialised lexicographic definitions, exploring denotation, connotation and conceptual structure. Georg Löckinger and Hendrik J. Kockaert examine intensional definitions and their integration with Unified Modeling Language (UML) diagrams, demonstrating how linguistic description and formal model‐ ling can complement each other. Questions of methodology and infrastructure are further explored by Tanja Wissik, who analyses the multidimensional nature of sustainability in termi‐ nology practice and highlights the role of research infrastructures, such as CLARIN and DARIAH, in which Budin himself has played a pivotal role. The international dimension of terminology development is addressed by Juan Carlos Díaz Vásquez and María Cecilia Plested Álvarez, who trace the evolution of terminology work in Colombia and illustrate how Budin’s approaches have been adapted in different contexts. Elena Chiocchetti and Natascia Ralli examine legal terminology in multilingual and system-bound contexts, while Frieda Steurs presents the work of the Centre of Expertise for Dutch Terminology, offering a concrete example of infrastructure building and collaborative practice. In the section on language technology and transdisciplinary linguistics, contributions reflect Budin’s sustained interest in multidimensional represen‐ 28 Dagmar Gromann, Barbara Heinisch & Vesna Lušicky <?page no="29"?> tations of linguistic and general knowledge and technological support of the language professions. Werner Winiwarter introduces a constructionist model for multilingual learning, while Dragoș Ciobanu analyses the impact of automation on the language industry. Jan Engberg investigates explanatory practices in communication about artificial intelligence, and Wolfgang U. Dressler explores text and discourse chaining as networks of meaning. The final section on translation addresses the changing professional lands‐ cape and the broader sociohistorical and technological contexts of translation. Peter Sandrini elaborates the concept of organisational translation compe‐ tence, Larisa Schippel analyses translation practices in historical and institu‐ tional settings, and Waltraud Kolb presents an empirical study of post-editing in literary translation. Xiaochun Zhang concludes this section with a study of multilingualism in video game localisation, highlighting the complexity of contemporary multilingual environments. As editors, we would like to conclude by offering Gerhard our warmest con‐ gratulations on the occasion of his 65 th birthday and his well-earned retirement. We express our deep gratitude for his remarkable scholarly contributions, his generosity as a mentor and colleague, and his enduring commitment to building bridges across disciplines and communities. It is both an honour and a pleasure to celebrate him with this volume. Bibliography Budin, Gerhard (1988). Möglichkeiten und Grenzen der Terminologieforschung in den Sozialwissenschaften. Vienna: University of Vienna. Budin, Gerhard (1993a). Wie (un)verständlich ist das Soziologendeutsch? : Begriffliche und textuelle Strukturen in den Sozialwissenschaften. Werkstattreihe Deutsch als Fremdsprache, Band-42. Frankfurt am Main, Wien: Peter Lang. Budin, Gerhard (1993b). Knowledge Organization and Modelling of Terminological Knowledge. In: Schmitz, Klaus-Dirk (ed.) Proceedings of Terminology and Enginee‐ ring (TKE’93). Frankfurt: Indeks Verlag, 1-7. Budin, Gerhard (1994). New challenges in specialized translation and technical commu‐ nication: An interdisciplinary outlook. In Snell-Hornby, Mary/ Pöchhacker, Franz/ Kaindl, Klaus (eds.). Translation Studies: An Interdiscipline. Amsterdam/ Philadelphia: John Benjamins, 247-254. Budin, Gerhard (ed.) (1995). Multilingualism in specialist communication. Proceedings of the 10th European LSP Symposium, Vienna, 29 Aug.-1 Sept. 1995. Vienna: Interna‐ tional Network for Terminology (IITF). Beyond Terminology: A tribute to Gerhard Budin 29 <?page no="30"?> Budin, Gerhard (1996a). Wissensorganisation und Terminologie: Komplexität und Dy‐ namik wissenschaftlicher Informations- und Kommunikationsprozesse. Forum für Fachsprachenforschung 28. Tübingen: Gunter Narr. Budin, Gerhard (1996b). Evolution of Scientific Terminologies. In: Galinski, Chris‐ tian/ Schmitz, Klaus-Dirk (eds.) Proceedings of Terminology and Knowledge Enginee‐ ring (TKE’96), 27-34. Indeks Verlag: Frankfurt. Budin, Gerhard/ Oeser, Erhard (eds.) (1997). Beiträge zur Terminologie und Wissenstech‐ nik. IITF Series, 7. Vienna: TermNet. Budin, Gerhard/ Melby, Alan K. (2000). Accessibility of Multilingual Terminological Resources - Current Problems and Prospects for the Future. In: Gavrilidou, Maria/ Ca‐ rayannis, George/ Markantonatou, Stella/ Piperidis, Stelios/ Stainhauer, Gregory (eds.) Proceedings of the Second International Conference on Language Resources and Evaluation (LREC’00), Athens, Greece. European Language Resources Association (ELRA). Budin, Gerhard (2001). A Critical Evaluation of the State-of-the-Art of Terminology Theory. Terminology Science and Research 12(1-2): 7-23. Budin, Gerhard (2006). Kommunikation in Netzwerken - Terminologiemanagement. In: Pellegrini, Tassilo/ Blumauer, Andreas (eds.). Semantic Web: Wege zur vernetzten Wissensgesellschaft. Berlin: Springer, 453-467. Budin, Gerhard (2007). L’apport de la philosophie autrichienne au développement de la théorie de la terminologie: ontologie, théories de la connaissance et de l’objet. In: Savatovsky, Dan/ Candel, Danielle (eds.). Genèse de la terminologie contemporaine (sources et réception). Langages, 168 (4), 11-23. Budin, Gerhard (2010a). Socio-terminology and computational terminology - toward an integrated, corpus-based research approach. In: De Cillia, Rudolf Gruber, Helmut/ Kry‐ zanowski, Michael/ Menz, Florian (eds.). Discourse, Politics, Identity, Stauffenburg Verlag: Tübingen, 21-31. Budin, Gerhard/ Wittenburg, Peter/ Romary, Laurent/ Declerck, Thierry (eds). (2010b). Language Resource and Language Technology Standards - state of the art, emerging needs, and future developments. Proceedings from a Workshop at the LREC Confe‐ rence 2010 in Malta. European Language Resources Association (ELRA). Budin, Gerhard/ Lušicky, Vesna (eds.) (2014). Languages for Special Purposes in a Multilingual, Transcultural World: Proceedings of the 19th European Symposium on Languages for Special Purposes, 8-10 July 2013, Vienna, Austria. University of Vienna: Vienna. Budin, Gerhard (2015). Digital Humanities, Language Industry, and Multilingualism: Global Networking and Innovation in Collaborative Methods. In Forstner, Mar‐ tin/ Lee-Jahnke, Hannelore (eds.). CIUTI-Forum 2014. Pooling Academic Excellence with Entrepreneurship for New Partnerships. Bern: Peter Lang, 423-448. 30 Dagmar Gromann, Barbara Heinisch & Vesna Lušicky <?page no="31"?> Budin, Gerhard (2016a). Der Beitrag der österreichischen Philosophie zur Entwicklung der Theorie der Terminologie. eDITion. Zeitschrift des Deutschen Terminologietages , 2016 (1), 5-16. Budin, Gerhard (2016b). Terminology as a key competence for translators. Journal of the Interpretation & Translation Institute, 20(4), 253-282. Budin, Gerhard (2016c). Arbeiten in der Sprachindustrie. In: Kadric, Mira/ Kaindl, Klaus (eds.). Berufsziel Übersetzen und Dolmetschen. Tübingen: Narr, 182-199. Budin, Gerhard (2018). Wüster, Schmitz und die Folgen - Entwicklungslinien der Termi‐ nologielehre und ihrer Anwendungsbereiche. In: Ahrens, Barbara/ Link, Lisa/ Schilly, Ute Barbara/ Wienen, Ursula (eds.). Verschmitzt! Von Terminologie und Terminologen. Berlin: Frank & Timme, 89-105. Budin, Gerhard (2019). Zum Entwicklungsstand der Terminologiewissenschaft. In: Pu‐ litano, Donatella/ Drewer, Petra (eds.). Terminologie: Epochen - Schwerpunkte - Umsetzungen. Zum 25-jährigen Bestehen des Rats für Deutschsprachige Terminolo‐ gie. Berlin: Springer, 7-19. Budin, Gerhard/ Elspaß, Stephan/ Lenz, Alexandra N./ Newerkla, Stefan M./ Ziegler, Arne (2019). The Research Project (SFB) ‘German in Austria’. Variation - Contact - Per‐ ception. In: Bülow, Lars/ Fischer, Ann Kathrin/ Herbert, Kristina (eds.). Dimensions of Linguistic Space: Variation - Multilingualism - Conceptualisations. Berlin: Peter Lang, 7-35. Felber, Helmut/ Budin, Gerhard (1989). Terminologie in Theorie und Praxis. Tübingen: Gunter Narr. Galinski, Christian/ Budin, Gerhard (1993). New trends in translation-oriented termino‐ logy management. In: Wright, Leland D./ Wright, Sue Ellen (eds.). Scientific and Technical Translation. Amsterdam/ Philadelphia: John Benjamins, 207-223. Haltmayer, Stepha/ Wuketits, Franz M./ Budin, Gerhard (eds.) (2000). Homo Pragma‐ tico-Theoreticus: Philosophie, Interdisziplinarität und Evolution. Erhard Oeser zum 60. Geburtstag. Wiener Arbeiten zur Philosophie, 4. Vienna: Peter Lang. Lukas, Helmut/ Stockinger, Johann/ Tjoa, A. Min/ Budin, Gerhard (1993). TONDONG: An Example of Multiperspective Representation of Ethnological Alliance Theory Using Hypermedia. In: Schmitz, Klaus-Dirk (ed.). Proceedings of Terminology and Engineering (TKE’93). Frankfurt: Indeks Verlag, 78-86. Lušicky, Vesna/ Budin, Gerhard (eds.) (2016). Multilingualism in Specialized Communi‐ cation: Challenges and Opportunities in the Digital Age. Proceedings of the 20th European Symposium on Languages for Special Purposes. University of Vienna: Vienna. Lušicky, Vesna/ Budin, Gerhard (2024). Terminologie und Sprachdaten im Lichte der Sprachenpolitik. In: de Cillia, Rudolf/ Reisigl, Martin/ Vetter, Eva (eds.). Sprachenpolitik in Österreich: Bestandsaufnahme 2021. Berlin, Boston: De Gruyter, 357-376. Beyond Terminology: A tribute to Gerhard Budin 31 <?page no="32"?> Rehm, Georg/ Marheinecke, Katrin/ Hegele, Stefanie/ Piperidis, Stelios/ Bontcheva, Ka‐ lina/ Hajič, Jan/ Choukri, Khalid/ Vasiļjevs, Andrejs/ Backfried, Gerhard/ Prinz, Chris‐ toph/ Gómez-Pérez, José Manuel/ Meertens, Luc/ Lukowicz, Paul/ van Genabith, Josef/ Lösch, Andrea/ Slusallek, Philipp/ Irgens, Morten/ Gatellier, Patrick/ Köhler, Joa‐ chim/ Le Bars, Laure/ Anastasiou, Dimitra/ Auksoriūtė, Albina/ Bel, Núria/ Branco, António/ Budin, Gerhard/ Daelemans, Walter/ De Smedt, Koenraad/ Garabík, Rado‐ van/ Gavriilidou, Maria/ Gromann, Dagmar/ Koeva, Svetla/ Krek, Simon/ Krstev, Cve‐ tana/ Lindén, Krister/ Magnini, Bernardo/ Odijk, Jan/ Ogrodniczuk, Maciej/ Rögnvalds‐ son, Eiríkur/ Rosner, Mike/ Pedersen, Bolette/ Skadiņa, Inguna/ Tadić, Marko/ Tufiș, Dan/ Váradi, Tamás/ Vider, Kadri/ Way, Andy/ Yvon, François (2020). The European language technology landscape in 2020: Language-centric and human-centric AI for cross-cultural communication in multilingual Europe. In: Calzolari, Nicoletta/ Bé‐ chet, Frédéric/ Blache, Philippe/ Choukri, Khalid/ Cieri, Christopher/ Declerck, Thierry/ Goggi, Sara/ Isahara, Hitoshi/ Maegaard, Bente/ Mariani, Joseph/ Mazo, Hélène/ Mo‐ reno, Asuncion/ Odijk, Jan/ Piperidis, Stelios (eds.). Proceedings of the 12th Language Resources and Evaluation Conference (LREC 2020). Paris: European Language Re‐ sources Association (ELRA), 3315-3325. Wright, Sue Ellen/ Budin, Gerhard (1994). Data elements in terminological entries: An empirical study. Terminology, 1(1), 41-59. Wright, Sue Ellen/ Budin, Gerhard (eds.) (1997). Handbook of Terminology Management. Volume I: Basic Aspects of Terminology Management. Amsterdam/ Philadelphia: John Benjamins. Wright, Sue Ellen/ Budin, Gerhard (eds.) (2001). Handbook of Terminology Management. Volume II: Application-Oriented Terminology Management. Amsterdam/ Philadel‐ phia: John Benjamins. Wüster, Eugen (1967). Wie die ISO-Empfehlung “Benennungsgrundsätze” entstanden ist - Professor Dr. Johannes Erich Heyde zum 75. Geburtstag am 22. Mai 1967. Muttersprache, 77.6, 169-180. Wüster, Eugen (1971). Begriffs- und Themaklassifikation. Unterschiede in ihrem Wesen und in ihrer Anwendung. Nachrichten und Dokumentation, 22.3, 98-104 und 22.4, 143-150. Wüster, Eugen (1974). Die Allgemeine Terminologielehre - ein Grenzgebiet Zwischen Sprachwissenschaft, Logik, Ontologie, Informatik und den Sachwissenschaften. Lin‐ guistics 12 (119), 61-106. 32 Dagmar Gromann, Barbara Heinisch & Vesna Lušicky <?page no="33"?> Begriffsbeziehungen Umsetzung in Termbanken und Austauschformaten Klaus-Dirk Schmitz Begriffe stehen nicht isoliert da, sondern bilden durch ihre Beziehungen zueinander ein System, in dem jeder Begriff seinen Platz und seine Bedeutung erhält. E U G E N W Ü S T E R (1898-1977) Abstract: Der Beitrag beleuchtet die Bedeutung von Begriffsbeziehungen in der Terminologiearbeit sowie deren Erfassung und Verwaltung in Ter‐ minologiedatenbanken (Termbanken). Neben hierarchischen Beziehungen - Abstraktions- und Bestandsbeziehungen - werden auch assoziative Begriffsbeziehungen betrachtet, deren Relevanz im Kontext von Ontolo‐ gien und Anwendungen der Künstlichen Intelligenz zunimmt. Auf Basis einschlägiger Normen wird gezeigt, dass bestehende Standards durchaus Unterstützung für eine adäquate Modellierung dieser Begriffsbeziehungen bieten. Zudem werden Ansätze vorgestellt, wie Begriffsbeziehungen in Termbanken strukturiert implementiert und im Terminologieaustausch, insbesondere mittels TBX, adäquat repräsentiert werden können. Dabei wird vor allem auf die TBX-Norm ISO 30042 und die Technische Spezifi‐ kation ISO/ TS 24624 eingegangen. 1 Einleitung Die Definition und systematische Strukturierung von Begriffen spielt eine zentrale Rolle in der Terminologiearbeit. Vor allem der Beziehung der Be‐ griffe untereinander kommt dabei eine besondere Bedeutung zu. Diese meist hierarchischen und im Kontext von Concept Maps und Ontologien auch oft assoziativen Begriffsbeziehungen müssen bei der praktischen Terminologie‐ <?page no="34"?> arbeit ermittelt, aufbereitet und in Terminologiedatenbanken (Termbanken) angemessen verwaltet werden. Dazu werden entsprechende Datenkategorien an den geeigneten Stellen des terminologischen Datenmodells implementiert, die dann in entsprechende (XML-)Konstrukte beim Austausch terminologischer Daten nach TBX überführt werden müssen. Die korrekte Umsetzung und Verwaltung solcher Begriffsbeziehungen sind entscheidend für die Konsistenz und Interoperabilität von terminologischen Daten. 2 Begriffsbeziehungen 2.1 Grundlagen In Rahmen der praktischen Terminologiearbeit werden Begriffe und ihre Be‐ nennungen in einer oder mehreren Sprachen untersucht. Die Ergebnisse werden in der Regel in begriffsorientierten Einträgen einer Terminologiedatenbank (Termbank) dokumentiert. Begriffe können zueinander in Beziehung gesetzt werden können, wobei meist ihre Merkmale eine wichtige Rolle spielen. Die Beziehungen zwischen Begriffen können hierarchischer oder nicht-hier‐ archischer Natur sein, wobei hierarchische Begriffsbeziehungen für die Termi‐ nologiearbeit bisher immer bedeutsamer waren als nicht-hierarchische. Aber im Kontext von Concept Maps, Ontologien und Anwendungen der Künstlichen Intelligenz gewinnen die nicht-hierarchischen Begriffsbeziehungen, die oft auch als assoziative Begriffsbeziehungen bezeichnet werden, immer mehr an Bedeutung. Bei den hierarchischen Begriffsbeziehungen wird zwischen Abstraktionsbe‐ ziehungen und Bestandsbeziehungen unterschieden. Unter den assoziativen Begriffsbeziehungen werden alle Arten zusammengefasst, die andere Aspekte der Beziehung abbilden, wie etwa sequentielle, chronologische oder kausale Beziehungen zwischen Begriffen. In diesem Beitrag soll nicht auf Begriffssysteme, Taxonomien, Klassifikati‐ onssysteme oder Ontologien eingegangen werden, die Beziehungen zwischen einer größeren Menge thematisch zusammenhängender Begriffe darstellen und in einem einheitlichen Ganzen ordnen. Vollständige und umfassende Begriffssysteme können zwar in Termbanken adäquat abgebildet und deren interne Begriffsbeziehungen durch eine Notation deutlich gemacht werden, werden aber bei der praktischen Terminologiearbeit in Unternehmen und Organisationen eher selten erarbeitet. 34 Klaus-Dirk Schmitz <?page no="35"?> 2.2 Arten von Begriffsbeziehungen 2.2.1 Abstraktionsbeziehungen Abstraktionsbeziehungen (auch: logische, generische oder taxonomische Bezie‐ hungen) orientieren sich an den Merkmalen der Begriffe. Sie sind dadurch gekennzeichnet, dass der Unterbegriff alle Merkmale seines Oberbegriffs enthält und gleichzeitig mindestens ein weiteres Merkmal besitzt. Die DIN 2342 (2022) definiert die Abstraktionsbeziehung als „hierarchische Begriffsbeziehung, bei der der Begriffsinhalt des untergeordneten Begriffs den Begriffsinhalt des übergeordneten Begriffs einschließt, wobei sich der untergeordnete Begriff in mindestens einem zusätzlichen Merkmal vom übergeordneten Begriff unter‐ scheidet“. Beispiel: Stuhl - Schreibtischstuhl Abstraktionsbeziehungen lassen sich oft leicht erkennen, wenn in Inhaltsdefi‐ nitionen der Begriff ausgehend von seinem Oberbegriff definiert wird. 2.2.2 Bestandsbeziehungen Bestandsbeziehungen (auch: Teil-Ganzes-Beziehungen oder partitive Beziehun‐ gen) zeichnen sich dadurch aus, dass der übergeordnete Begriff (gedanklich) in seine einzelnen Teilbegriffe zerlegt werden kann. Sie beruhen damit auf der Beziehung des Ganzen zu seinen Teilen. Das Ganze, also der übergeordnete Begriff, wird Verbandsbegriff genannt, der untergeordnete heißt Teilbegriff. Die DIN 2342 (2022) definiert die Bestandsbeziehung als „hierarchische Begriffsbeziehung, bei der der übergeordnete Begriff sich auf einen Gegenstand als Ganzes bezieht und der untergeordnete Begriff sich auf einen Teil dieses Gegenstands bezieht“. Beispiel: Stuhl - Rückenlehne 2.2.3 Assoziative Begriffsbeziehungen Assoziative Begriffsbeziehungen (auch: pragmatische oder einfach nicht-hier‐ archische Beziehungen) beruhen nicht auf hierarchischen, sondern auf anderen thematischen Zusammenhängen zwischen Begriffen. Oft werden die assoziativen Begriffsbeziehungen noch in sequenzielle, tem‐ porale, kausale oder oppositionelle Beziehungen unterteilt, so in einer früheren Fassung der DIN 2342, aber diese Unterteilung kann nicht umfassend sein und wurde deswegen in der aktuellen Fassung der Norm aufgegeben. Dagegen versucht die aktuelle Fassung der ISO 704 (2022), mehr als 30 unterschiedliche Arten von assoziativen Begriffsbeziehungen zu unterscheiden. Begriffsbeziehungen 35 <?page no="36"?> Beispiel: Stuhl - Nussholz (Relation „besteht aus“) Assoziative Begriffsbeziehungen sind in der Terminologiewissenschaft lange vernachlässigt worden, sind aber heute im Kontext von Concept Maps, Wis‐ sensgraphen und Ontologien wichtig und werden, je nach Anwendungsbereich und Interesse, ganz unterschiedlich definiert und genutzt. 3 Begriffsbeziehungen in Termbanken 3.1 Grundlagen Es ist ein Verdienst der Normung, aber auch einschlägiger Literatur (Arntz et al. 2021; Drewer und Schmitz 2017; DTT 2014) und Schulungen, dass Terminologie heute nach anerkannten und weitgehend einheitlichen Standards erarbeitet und verwaltet wird. So sind Begriffsorientierung und Benennungsautonomie durch das terminologische Metamodell (z. B. aus DIN ISO 26162-1 2020: 12) vorgegeben, und auch die bei der Modellierung von Termbanken verwendeten Datenkategorien orientieren sich an der ISO 12620-1 (2022) sowie dem Datenka‐ tegorienverzeichnis (DatCatInfo o. J.). Beide Standards bieten jedoch nur wenig Hilfestellung für eine einheitliche Verwaltung von Begriffsbeziehungen und Begriffssystemen in Termbanken. Im Folgenden werden Möglichkeiten aufgezeigt, wie Beziehungen zwischen Begriffseinträgen in der Terminologieverwaltung realisiert werden können und wie diese Informationen beim Terminologieaustausch (mit TBX) weitergegeben werden können. 3.2 Implizite Begriffsbeziehungen In der Praxis werden Beziehungen zwischen Begriffen in vielen Terminologie‐ verwaltungssystemen oft sehr unspezifisch durch Verweise von einem Eintrag auf einen anderen umgesetzt. Dabei kann zwischen impliziten und expliziten Verweisen unterschieden werden. Bei impliziten Verweisen wird entweder in einer eigenen Datenkategorie (z. B. Siehe oder Siehe auch) oder in einer anderen, bereits vorhandenen Datenkategorie (z. B. Definition) eine Benennung ausgewählt und „dahinter“ ein Verweis auf einen anderen Eintrag angelegt (siehe Abb. 1). Die Verlinkung im Hintergrund sollte immer über eine eindeu‐ tige Begriffs-/ Eintragsnummer und nicht über die Benennung erfolgen, da es sonst bei Unklarheiten/ Mehrdeutigkeiten oder Benennungsänderungen zu Problemen kommen kann. 36 Klaus-Dirk Schmitz <?page no="37"?> Abb. 1: Implizite Begriffsverknüpfungen in MultiTerm (Trados) Auch wenn Datenkategorien für solche impliziten Verweise zwischen Begriffs‐ einträgen in Termbanken durchaus üblich sind, insbesondere für Verweise von Inhaltsdefinitionen auf den jeweiligen Oberbegriff (siehe Definition in Abb. 1), und auch wenn es standardisierte terminologische Datenkategorien wie <see> oder <see also> gibt, sind implizite Begriffsbeziehungen schwer zu pflegen und konsistent zu halten. Außerdem erlauben sie es nicht, mehr als zwei Begriffe „auf einen Blick“ zu erfassen, wie es bei größeren Terminologieverwaltungssystemen der Fall ist. Ein weiterer Nachteil ist, dass derartige implizite Begriffsverknüp‐ fungen in der Regel auf der Sprachebene - und bei Kontexten sogar auf der Benennungsebene - des terminologischen Eintrags erfolgen: sollen und können diese impliziten Verknüpfungen bei mehrsprachigen Einträgen in allen Sprachen angemessen verwaltet werden? 3.3 Explizite Begriffsbeziehungen In vielen Terminologieverwaltungssystemen können Begriffsbeziehungen aber auch explizit über spezifische Datenkategorien verwaltet werden. Hierfür bieten sich in Anlehnung an die ISO 12620 (1999) bzw. das Data Category Repository (DatCatInfo o. J.) verschiedene Realisierungsmöglichkeiten an, da sowohl die Art der Begriffsbeziehung als auch die hierarchische Position angegeben und zusätzlich die Referenz untergebracht werden muss (siehe auch DIN ISO 26162-1 2020). Für die Realisierung eines Verweises auf einen Oberbegriff, der in einer Abstraktionsbeziehung (generische Beziehung) zum aktuellen Begriff steht, gibt es z. B. folgende 3 Möglichkeiten: 1. <Generischer Oberbegriff> (Verweis) - 2. <Oberbegriff> (Verweis) <Art> generisch 3. <Generische Beziehung> (Verweis) <Richtung> Oberbegriff Abbildung 2 zeigt beispielhaft die Realisierung von Möglichkeit 2 in MultiTerm, wobei die Datenkategorien aus Vereinfachungsgründen nicht ganz normgerecht Begriffsbeziehungen 37 <?page no="38"?> benannt wurden (Unterbegriff bei partitiver Begriffsbeziehung sollte eigentlich Teilbegriff heißen). Die entsprechenden Metadaten für die Begriffsbeziehungen sind auf Eintrags-/ Begriffsebene (oben) im terminologischen Eintrag angeord‐ net; die Verweise sind über die deutsche Benennung sichtbar, im Hintergrund aber sprachunabhängig über die Begriffsnummer realisiert. Denkbar wäre auch, die Begriffsbeziehungen in allen Sprachen auf den jeweiligen Sprachebenen anzugeben, was zur Benutzerfreundlichkeit beiträgt, aber den Aufwand bei der Eintragserstellung erhöht und die Gefahr von Inkonsistenzen in sich birgt. Abb. 2: Explizite Begriffsverknüpfungen in MultiTerm (Trados) Durch diese explizite Art der Begriffsverknüpfung entsteht innerhalb einer Termbank (für ein bestimmtes Fachgebiet) ein Geflecht von Relationen zwischen terminologischen Einträgen, das recht komplex werden kann und einen hohen Aufwand zur Aufrechterhaltung der Konsistenz erfordert. Zudem werden diese Begriffsverknüpfungen in der Regel nicht reziprok angelegt und verwaltet, d. h. ein Verweis vom aktuellen Begriff auf den Oberbegriff führt nicht automatisch zu einem (Rück-)Verweis vom Oberbegriff auf den aktuellen (Unter-)Begriff. Falls gewünscht, müsste auch dieser Verweis manuell angelegt werden. 38 Klaus-Dirk Schmitz <?page no="39"?> 3.4 Autonome Begriffsbeziehungen Ähnlich wie explizite Begriffsbeziehungen werden auch autonome Begriffsbe‐ ziehungen realisiert: Im terminologischen Eintrag oder auch außerhalb des terminologischen Eintrags werden Begriffsbeziehungen meist in der Form <Begriff 1> - <Relation> - <Begriff 2> angegeben. Die Relation selbst kann hierarchisch (Abstraktionsrelation oder Bestandsrelation) oder nichthierarchisch und unidirektional oder bidirektional sein. Insbesondere bei nichthierarchischen Begriffsbeziehungen sind häufig weitere Angaben möglich, um z. B. die Art der Beziehung näher zu charakte‐ risieren (z. B. sequentielle, gegensätzliche, pragmatische oder sonstige Bezie‐ hung). Häufig können solche Begriffsbeziehungen bei der Modellierung der Termbank selbst definiert werden, um unternehmens-, anwendungs- oder fachspezifische Relationen wie „hilft bei“, „wird hergestellt von“ oder „steuert“ verwalten zu können (siehe Abb. 3). Abb. 3: Festlegung von Begriffsbeziehungen in LookUp (D.O.G.) Mit dieser autonomen Verwaltung von Begriffsbeziehungen und der eigen‐ ständigen Definition von (nicht-hierarchischen) Begriffsbeziehungen ist der Übergang von Begriffssystemen zu Ontologien oder Taxonomien geebnet. Häufig lässt sich das Geflecht dieser autonomen Begriffsbeziehungen grafisch als Concept-Map darstellen und Begriffe bzw. Begriffsbeziehungen können sogar auf der grafischen Oberfläche manipuliert werden (siehe Abb. 4). Die Beschriftung der Knoten und teilweise auch der Relationen kann bei mehrspra‐ chigen Termbanken - je nach Tool - durch Wechsel der Quell-/ Suchsprache geändert und an die Sprache des Benutzers angepasst werden. Concept Maps können damit sowohl an den Begriffsknoten als auch an den Relationen mit Benennungen in jeder in der Termbank verwalteten Sprache beschriftet werden. Begriffsbeziehungen 39 <?page no="40"?> 1 www.coreon.com 2 www.dog-gmbh.de 3 www.interverbumtech.com 4 www.kaleidoscope.at Abb. 4: Darstellung von Begriffsbeziehungen als Concept-Map in LookUp (D.O.G.) Die Möglichkeit zur Verwaltung, grafischen Darstellung sowie Änderung von autonomen, hierarchischen und nicht-hierarchischen Begriffsbeziehungen und deren Verknüpfung mit den terminologischen Einträgen besteht bei verschie‐ denen Software-Lösungen, wie beispielsweise bei Coreon 1 , LookUp 2 , TermWeb 3 und Kalcium-Quickterm 4 . 40 Klaus-Dirk Schmitz <?page no="41"?> 5 www.tbxinfo.net 4 Begriffsbeziehungen in TBX Für die Repräsentation und Weitergabe terminologischer Einträge wird das XML-basierte Austauschformat für terminologische Datenbestände, das als TBX bezeichnet wird, einer eingehenden Betrachtung unterzogen. Die TBX-Website 5 bietet interessierten Personen freien Zugang zu Beispielen, Anwendungen und weiteren Informationen zu TBX. Obwohl die TBX-Norm von 2019 bereits Möglichkeiten beschreibt, implizite und explizite Begriffsbeziehungen darzu‐ stellen und damit beim Terminologieaustausch zu übermitteln, bestand doch der Bedarf, gerade explizite Begriffsbeziehungen eindeutiger zu spezifizieren. Aus diesem Grund wurde im Jahr 2021 die Technische Spezifikation ISO/ TS 24634 (2024) publiziert und 2024 aktualisiert, die sich explizit mit der Repräsentation von Begriffsbeziehungen und Fachgebietsangaben in TBX befasst und somit ISO 30042 (2019) ergänzt. Gemäß ISO/ TS 24634 (2024) werden die Abstraktionsbeziehung (generic) und die Bestandsbeziehung (partitive) als hierarchische Begriffsbeziehungen sowie eine nichthierarchische Beziehung (associative) definiert, die einer weiteren Spezifizierung unterliegt. Die Beziehungen werden auf der Ebene der Begriffe implementiert, wobei die type-Elemente die Relation und die target-Elemente die Begriffsnummer des Zielbegriffs repräsentieren. Zwischen Start- und End-Tag steht die (Vorzugs-)Benennung des Zielbegriffs (siehe Abb. 5). Es existieren fol‐ gende Möglichkeiten bezüglich der Art der Beziehung zwischen den Begriffen, die durch sogenannte type-Werte spezifiziert werden können: - genericConcept - specificConcept - comprehensiveConcept - partitiveConcept - associativeConcept Begriffsbeziehungen 41 <?page no="42"?> Abb. 5: Wechselseitige Abstraktionsbeziehung in TBX Will man bei nichthierarchischen Begriffsbeziehungen (associativeConcept) den Beziehungstyp weiter verfeinern, so bietet sich hierfür das Element <desc‐ ripNote> in TBX an. Dies kann über zwei „Verfeinerungsstufen“ umgesetzt werden (siehe Abb. 7). ISO/ TS 24634 (2024) enthält in Anhang A eine Liste von nichthierarchischen Begriffsbeziehungen aus der ISO 704 (2022) mit 8 Gruppen und 32 Untergruppen; Abbildung 6 zeigt einen Auszug daraus. Abb. 6: Auszug aus ISO/ TS 24634: 2024, Anhang A 42 Klaus-Dirk Schmitz <?page no="43"?> Abb. 7: Zweistufig spezifizierte nichthierarchische Begriffsbeziehung in TBX Die in Abbildung 7 dargestellte Lösung findet nicht nur Anwendung bei den nichthierarchischen Begriffsbeziehungen aus dem Anhang A von ISO/ TS 24634 (2024), sondern auch bei anderen anwendungs- oder unternehmensspezifischen nicht hierarchische Begriffsbeziehungen. Dies umfasst sowohl einstufige (nur relationType) als auch zweistufige (relationType und relationRole) Typologien. Die Norm ISO 30042 (2019) spezifiziert die Darstellung von internen und insbesondere externen Begriffsbeziehungen im terminologischen Austauschfor‐ mat TBX in praktikabler Form. Um auch eine Lösung für autonome Begriffs‐ beziehungen in TBX bereitzustellen, wurde die Vorgängerfassung der ISO/ TS 24634 (2021) aktualisiert; darin wird folgende Umsetzung festgelegt (siehe Abb. 8). Begriffsbeziehungen 43 <?page no="44"?> Abb. 8: Definition und Nutzung einer autonomen Begriffsbeziehung in TBX Die autonomen Begriffsbeziehungen werden in TBX nicht in den eigentlichen begriffsorientierten Einträgen (<conceptEntry>) und damit auch nicht im <body> der TBX-Instanz, sondern im Anhang (<back>) der TBX-Instanz anlegt. Nehmen wir an, es sollte eine autonome Begriffsbeziehung zwischen dem ter‐ minologischen Eintrag für „KI-Chatbot“ (ID 10239) und dem terminologischen Eintrag „Terminologieextraktion“ (ID 10438) in TBX dargestellt werden, die auf der selbst definierten assoziativen Begriffsbeziehung „benutzt für“ basiert und die Aussage „KI-Chatbot wird für Terminologieextraktion benutzt“ abbildet. Im unteren Teil der Abbildung 8 wird diese eigene Begriffsbeziehung spezifiziert und im oberen Teil der Abbildung 8 als autonome Begriffsbeziehung zusammen mit den beiden relevanten Begriffen eingesetzt. Sowohl die autonome Begriffs‐ beziehung als auch die Definition der anwendungs- oder nutzerspezifischen assoziativen Begriffsbeziehung wird im <back> der TBX-Instanz spezifiziert und für den Austausch eindeutig beschrieben. Diese Lösung zur Repräsentation autonomer Begriffsbeziehungen in TBX hat den großen Vorteil für Terminologieverwaltungssysteme, die autonome Begriffsbeziehungen verwalten, dass die Begriffsbeziehungen beim Datenexport nach TBX nicht von der autonomen Verwaltung in explizite Begriffsbeziehun‐ gen in jedem Eintrag - und evtl. auch in jeder Sprache - umgewandt werden müssen. Ebenso müssen beim Import terminologischer Daten aus TBX-Instan‐ 44 Klaus-Dirk Schmitz <?page no="45"?> zen nicht die expliziten Begriffsbeziehungen innerhalb der einzelnen termino‐ logischen Einträge in eine autonome Begriffsbeziehungsrepräsentation umge‐ wandt werden. 5 Fazit Die systematische Erfassung und Verwaltung von Begriffsbeziehungen bildet die Grundlage für eine konsistente und interoperable Terminologiearbeit. Nur wenn hierarchische und assoziative Beziehungen zwischen Begriffen klar de‐ finiert, korrekt modelliert und technisch sauber in Terminologiedatenbanken sowie Austauschformaten wie TBX umgesetzt werden, kann eine einheitliche und verlässliche Nutzung terminologischer Daten gewährleistet werden. Während früher - zumindest in klar definierten Fachgebieten - der Schwer‐ punkt eher auf den hierarchischen Begriffsbeziehungen und dem Ausbau vollständiger Begriffssysteme oder umfassender Taxonomien lag, geht heute die Tendenz zur Kombination von hierarchischen und assoziativen Begriffsbezie‐ hungen im Rahmen von Wissensnetzen und Ontologien. Diese helfen nicht nur Anwendungen der Künstlichen Intelligenz (KI) in der Aufbereitung von Wissen, auch eine KI-Unterstützung bei der Identifizierung und Herausarbeitung von Beziehungen zwischen Begriffen kann sehr hilfreich sein. Bibliographie Arntz, Reiner/ Picht, Heribert/ Schmitz, Klaus-Dirk (2021): Einführung in die Terminolo‐ giearbeit. 8. Aufl. (unveränderter Nachdruck der 7., vollständig überarbeiteten und aktualisierten Auflage). Hildesheim: Georg Olms Verlag. DatCatInfo (o. J.). Data Category Repository. Abrufbar unter: https: / / datcatinfo.termwe b.eu/ client/ twd (Stand: 28.04.2025) DIN 2330 (2022). Terminologiearbeit - Grundsätze und Methoden. Berlin: Beuth. DIN 2331 (2019). Begriffssysteme und ihre Darstellung. Berlin: Beuth. DIN 2342 (2022). Terminologiewissenschaft und Terminologiearbeit - Begriffe. Berlin: Beuth. DIN ISO 26162-1 (2020). Management von Terminologieressourcen - Terminologieda‐ tenbanken - Teil 1: Design. Berlin: Beuth. Drewer, Petra/ Schmitz, Klaus-Dirk (2017). Terminologiemanagement: Grundlagen - Methoden - Werkzeuge. Wiesbaden: Springer Vieweg. Drewer, Petra/ Schmitz, Klaus-Dirk (2017). Begriffsbeziehungen - Terminologische Neu‐ igkeiten aus DIN und ISO. In: Drewer, Petra/ Mayer, Felix/ Pulitano, Donatella (Hrsg.). Terminologie: Tools und Technologien. Köln: Deutscher Terminologie-Tag, 47-62. Begriffsbeziehungen 45 <?page no="46"?> DTT - Deutscher Terminologie-Tag (2014). Terminologiearbeit - Best Practices 2.0. 2. Koordination und Redaktion: Drewer, Petra; Pulitano, Donatella; Schmitz, Klaus-Dirk. Köln: Deutscher Terminologie-Tag e.V. ISO 704 (2022). Terminology work — Principles and methods. Genf: International Organization for Standardization. ISO 12620 (1999). Computer applications in terminology — Data categories. Genf: International Organization for Standardization. ISO 12620-1 (2022). Management of terminology resources — Data categories — Part 1: Specifications. Genf: International Organization for Standardization. ISO/ TS 24634 (2024). Management of terminology resources — TermBase eXchange TBX-compliant representation of concept relations and subject fields. Genf: Interna‐ tional Organization for Standardization. ISO 30042 (2019). Management of terminology resources - TermBase eXchange (TBX). Genf: International Organization for Standardization. Schmitz, Klaus-Dirk (2022). Normen für Terminologiearbeit, Technische Redaktion und Übersetzen. 2., überarbeitete und erweiterte Auflage. Berlin: Beuth. 46 Klaus-Dirk Schmitz <?page no="47"?> 1 Acknowledgements: This work is supported by the HEREDITARY Project, as part of the European Union’s Horizon Europe research and innovation programme under grant agreement No GA 101137074. This work is also part of the initiatives carried out the by the Portuguese national funding through the FCT - Portuguese Foundation for Science and Technology, I.P. as part of the project UID/ 03213/ 2025 - Linguistics Research Centre of NOVA University Lisbon (CLUNL). https: / / doi.org/ 10.54499/ UID/ 03213/ 2025. Terminological and Specialised Lexicographic Definitions: Navigating Connotation and Denotation 1 Rute Costa Abstract: This article examines the distinctions between terminological and specialised lexicographic definitions in specialised domains. Although both represent and communicate knowledge, they differ in purpose, struc‐ ture, and audience. Termniological definitions establish precise relations‐ hips between terms and concepts by identifying essential characteristics through denotative methods. They support conceptual clarity, knowledge organisation, and interoperability across linguistic and computational systems. In contrast, specialised lexicographic definitions prioritise acces‐ sibility and contextual understanding for users. They rely on natural language and include denotative and connotative aspects, incorporating cultural and pragmatic nuances. Drawing on Wüster’s framework and ISO/ TC37, this study analyses how denotation and connotation shape definitional practices, offering a structured comparison and outlining implications for specialised communication and resource development. 1 Introduction This chapter examines the distinction between terminological and specialised lexicographic definitions, highlighting their roles and characteristics within specialised domains. While both types of definitions serve to represent and communicate knowledge, they differ in purpose, structure, and audience. <?page no="48"?> Understanding these differences is essential for developing accurate linguistic and terminological resources. Terminology focuses on the systematic organisation of domain-specific knowledge, aiming to establish a precise and stable relationship between terms and their corresponding concepts. Following ISO 704 (2022) and ISO 1087 (2019), terminological definitions identify the essential characteristics of a concept, using primarily denotative methods. They may take formal, semi-formal, or verbal forms depending on the required level of conceptual precision and the needs of the target user group, supporting coherent communication and consistent knowledge management within specialised communities. Specialised lexicographic definitions, by contrast, address a broader audience and prioritise usability and contextual understanding. They rely solely on na‐ tural language and capture both denotative and connotative aspects, including social, cultural, and pragmatic nuances that influence how terms are interpreted in realworld communication. This chapter adopts a theoretical approach, drawing on Wüster’s framework and subsequent developments in ISO/ TC37 standards, as well as studies by Costa (2013) and Santos and Costa (2015). It explores how the interaction between the goals of a resource, the emphasis on denotation, and the context-sensitive, interpretive nature of connotation shapes the construction of definitions in different types of specialised resources. The structure of this chapter is as follows. Section 2 presents the theoretical framework underpinning the treatment of definitions in both terminology science and specialised lexicography. Section 3 engages with the debates sur‐ rounding the concepts of denotation and connotation, which are analysed in a contrastive manner to clarify their respective functions in definitions. Section 4 presents a structured differentiation between terminological and specialised lexicographic definitions. Finally, Section 5 provides illustrative examples, followed by concluding remarks that synthesise the principal insights and implications derived from the chapter’s discussion. 2 Linking Terminology and Lexicography through Definition The intersection of terminology and specialised lexicography reveals both complementary and divergent approaches to the treatment of definitions. While each discipline operates within its own methodological and theoretical traditions, the definition functions as a central point of contact. This section explores how definitions operate at the interface between terminological ac‐ curacy and lexicographic descriptiveness. By examining the ways in which 48 Rute Costa <?page no="49"?> each field conceptualises and formulates definitions, we aim to highlight the shared objectives, distinct constraints, and potential areas of convergence. Particular attention is given to how definitions mediate between concepts and terms in terminology, versus how they reflect language use and variation in lexicography. This comparative perspective offers a deeper understanding of how both disciplines complement each other in contributing to the organisation, dissemination, and sharing of specialised knowledge. 2.1 Terminology and Definition Terminology is a twofold scientific discipline (Costa 2013; Santos and Costa, 2015) rooted in epistemological theories of knowledge, concept and term formation (Budin 2007, 2022; Wüster 1979; Felber 1984; Lauren et al. 1998; Rey, 1979; Savatovsky 2022). It engages, on the one hand, with definitions as epistemic tools for delimiting, identifying, and situating concepts within coherent systems of specialised knowledge, and on the other hand, with terms as linguistic instantiations of concepts, which function as observable markers of conceptual existence and play a constitutive role in the production, articulation, and transmission of specialised knowledge. Definitions help us structure and communicate knowledge; they are instruments through which knowledge becomes shareable within a community. The concept of definition is defined by ISO 1087 (2019) as the “representation of a concept by an expression that describes it and differentiates it from related concepts.” In this definition, “expression” refers not only to a linguistic statement but also to a formal representation. A formal representation can be considered a definition as long as it clearly represents a concept, identifies its characteristics, and distinguishes it from related concepts. Linguistic definitions, commonly used in human communication and traditional terminology work, should ideally reflect the structure and precision of the formal representation. Definitions play a fundamental role in the formalisation of knowledge representation. By specifying the characteristics of concepts in a precise and consistent manner, definitions help eliminate ambiguity and ensure shared understanding within a community. This formalisation process aims to make knowledge systematic and unambiguous, often in ways that are suitable for logical reasoning or computational processing. Knowledge representation, particularly in disciplines such as philosophy, linguistics, artificial intelligence, and cognitive science, relies heavily on well-crafted definitions to structure, store, and communicate information effectively. In terminology work, this Terminological and Specialised Lexicographic Definitions 49 <?page no="50"?> 2 https: / / www.w3.org/ OWL formal approach ensures that definitions function as essential tools to represent concepts in a coherent, interoperable, and shareable way. The terminologist renders formal representations into natural language definitions, recognising that while these definitions adhere to the underlying logic of the discourse, they remain open to interpretation, making it challenging to eliminate ambiguity. We consider linguistic and formal definitions complementary, as they serve distinct purposes: natural language definitions support human understanding, whereas formal definitions, grounded in logic or expressed in OWL 2 , enable machine processing. Together, they bridge human communication and compu‐ tational applications, enhancing knowledge sharing, information retrieval, and automated reasoning. Leveraging both allows for a more effective and versatile management of concepts. Terminology work defined by ISO 1087 (2019) as “work concerned with the systematic collection, description, processing, and presentation of concepts,” relies on a structured, methodical approach, with concept definition at its core. Concepts exist within concept systems, “a set of concepts structured in one or more related domains according to the concept relations among its concepts”, where analysing their characteristics is essential for distinguishing related concepts. 2.2 Specialised Lexicography and Definition In contrast, specialised lexicography focuses on specialised lexical units (terms) and their usage across a variety of communicative situations, reflecting both linguistic function and social pragmatics. Unlike general lexicography, which addresses everyday language, specialised lexicography deals with domain-spe‐ cific vocabulary and draws on both lexicographic and terminological principles. While both disciplines focus on terms and terminological information, con‐ temporary terminology, especially in computational contexts, tends to be more machine-oriented, designed for automated processing rather than casual human reading, whereas lexicography is more human-oriented. This distinction shapes how information is organised and managed. Costa (2013) considers the two approaches as complementary, corresponding to two different social needs, the first prioritising concept analysis and how concepts are designated and reflected by language, the second specialised language use. 50 Rute Costa <?page no="51"?> Specialised lexicography produces resources such as specialised dictionaries, glossaries, and databases aimed at translators, students, and other non-expert users engaged with specialised knowledge and communication. As defined by Humbley (2017: 317), specialised dictionaries are as follows: […] specialised nature of the subjects they treat, focusing on particular subject fields, professional practices or even leisure activities such as sports with the aim of helping the user to acquire specialised knowledge, read specialised texts or engage in specialised translation. We consider these resources hybrid, as they focus on terms and occupy a posi‐ tion between lexicographic and terminological resources, drawing upon both lexicographic and terminological theoretical and methodological frameworks. Specialised lexicography responds to the need for resources that not only offer accessible terms, definitions, descriptions and explanations for a broader range of users, but also provide detailed information on term usage, including collocations, examples, and contextual explanations. By bridging the gap bet‐ ween terminology and everyday language, specialised lexicography helps users better understand and apply domain-specific terms in real-world contexts. These differences stem from the distinct social and pragmatic functions the resources fulfil. In contrast, terminological resources are typically designed for domain experts who require clarity and precision in professional communica‐ tion, in order to avoid misunderstandings, ensure accurate communication, sup‐ port professional documentation, and enable reasoning and decision-making. Moreover, their structured and hierarchical nature makes them particularly well-suited for machine-readable applications, such as information retrieval, natural language processing, knowledge representation systems, and ontolo‐ gies, where unambiguous and consistently defined concepts are essential for accurate and efficient automated processing. 3 The Terminological and Lexicographic Definitions: The Significance of Denotation and Connotation Denotation and connotation form a dichotomous pair, making it difficult to address one without reference to the other. The notion of connotation has been extensively theorised. (Bloomfield 1933; Gary-Prieur 1971; Kerbrat-Orecchioni 1977; Martinet 1969; Molino 1971; Costa and Vezzani 2024). Bloomfield high‐ lighted connotation as a significant component of meaning, influencing both terminological and lexicographic theory today. The roles of denotation and con‐ notation remain debated within terminology studies: the Wüsterian tradition Terminological and Specialised Lexicographic Definitions 51 <?page no="52"?> seeks to eliminate connotative elements in favour of objective, comprehensive definitions, whereas communicative (Cabré 1999), socioterminological (Gaudin 1993), and sociocognitive (Temmerman 2020) frameworks argue that purely denotative definitions cannot capture language’s dynamic, context-dependent nature of real language use. Gary-Prieur (1971) characterises connotation as the additional, plural mea‐ nings of a sign arising from individual and contextual use, with each denotation linked to an open-ended set of connotations. Similarly, Nilsson (2024) defines connotation as subjective meaning that supplements basic signification in communication. While denotation provides a stable conceptual anchor, especi‐ ally in scientific and technical language, connotation conveys contextual and emotional dimensions. In terminology, where precision is crucial, connotation can introduce distortion, yet its variability also underpins the adaptability of natural language, sustaining ongoing debates in terminology and specialised lexicography. Denotation, described by Gary-Prieur (1971) as a sign’s reference to an extralinguistic object and by Lyons (1995) as the relation between expressions and their referents, constitutes the shared core meaning within a linguistic community (Cruse 2000). Terminology prioritises denotation to establish precise, context-independent concepts, emphasising essential characteristics and excluding subjective ele‐ ments to ensure clarity and consistency in expert communication. Specialised lexicography integrates both denotation and connotation. While acknowledging core referential meaning, it also incorporates social, cultural, emotional, and pragmatic nuances that shape a term’s use in context. By accounting for variation in register, user background, polysemy, metaphor, and evaluative overtones, it offers context-sensitive descriptions suited to translators, language learners, and general users. This contrast reflects their distinct functions: terminology prioritises denota‐ tive precision for expert communication and knowledge management, whereas specialised lexicography includes connotation to support broader comprehen‐ sion and culturally informed use across contexts. 4 Terminological Definition and Specialised Definition Terminological definitions are generally intended for domain experts, with the primary objective of establishing precise and unambiguous conceptual boundaries. Depending on the theoretical framework adopted and the intended social application, it becomes necessary to write not only definitions oriented 52 Rute Costa <?page no="53"?> toward domain-specific expertise, but also more descriptive, connotative and context-sensitive definitions designed to meet non-experts or experts from other fields. This is why it is important to acknowledge the diversity of definition types used in terminology work, each serving different needs and contexts: Fig. 1: Types of definitions When dealing with terms and concepts, it is necessary to employ various types of definitions tailored to different purposes and contexts. For instance, some definitions may aim to provide broad linguistic statements suitable for general comprehension, as it is typical of lexicographic definitions. In contrast, others require precise and formal representations to support specialised com‐ munication within a particular disciplinary context, these are referred to as specialised lexicographic definitions. Accordingly, terminological work must carefully consider the nature of the concept, the intended audience, and the domain-specific requirements to determine the most appropriate definitional approach. As illustrated in Figure 1, a terminological definition is a specific type of definition that follows strict principles of conceptual clarity, coherence, and domain relevance. They aim is to reflect the structured organisation of knowledge within a specific field, often aligning with formal ontologies or concept systems. In contrast, more general definitions (including those found in specialised dictionaries) are typically designed for broader audiences and may prioritise accessibility over precision. Terminological and Specialised Lexicographic Definitions 53 <?page no="54"?> ISO TC 37’s framework for definitions, rooted in the Aristotelian tradition and influenced by Frege (1971) and Carnap (Budin 2022), emphasises formal semantics and logical clarity. In ISO 1087 (2019), definitions implicitly serve terminological purposes. The framework recognises several types - intensional definitions (essential characteristics), extensional definitions (enumerating sub‐ ordinate concepts), and generic or partitive extensional definitions (based on hierarchical or partitive relations) - all crucial for precise concept characterisa‐ tion within a concept system. Fig. 2: Types of terminological definitions based on ISO 1087 (2019) Terminological definitions aim to establish clear conceptual boundaries by sys‐ tematically specifying a concept’s essential characteristics. A common approach is the genus-differentia model, which defines a concept by first identifying its broader category (genus) and then detailing the distinguishing features (differentiae) that set it apart from related concepts. This structure is widely used to clarify both the hierarchy and the distinctiveness of concepts. Whether a definition seeks to clarify a concept (conceptual focus) or stabilise meaning (linguistic focus), the method is often similar: it typically involves identifying a genus and a specific difference. As Dotoli and Boccuzi (2012: 35) 54 Rute Costa <?page no="55"?> 3 “The first defines belonging; the second defines difference. The word either is or is not.” (Author translation) 4 The authors talk about “lexie”. 5 https: / / bioportal.bioontology.org/ ontologies/ MESH 6 Research Portofolio Online Reporting Tool - https: / / web.archive.org/ web/ 2009062201 1306/ http: / / projectreporter.nih.gov/ reporter.cfm 7 https: / / www.merriam-webster.com/ medical put it, “Le premier définit l’appartenance, le deuxième la différence. Le mot est ou n’est pas.” 3 By extension, we might say: the concept either is or is not. We regard terminology and specialised lexicography as distinct disciplines, each with its own objectives, methodologies and outcomes. Together, they contribute to a more refined understanding of the relationship between con‐ ceptual and lexical structures, a view we have advanced in prior work (Costa 2003). Although the statement by Mel’čuk and Polguère (2016: 61) is rooted in Explanatory and Combinatorial Lexicology, a component of the Meaning-Text Theory, we share their definition of a lexicographic definition, namely, that its purpose is not to represent the entity or the fact denoted by the lexical unit being defined. 4 5 Examples In this final section, we will illustrate the differences between a terminological definition and a specialised lexicographic definition. To do so, let us consider the concept <brain-gut axis>, <Alzheimer’s disease> and <mucosal immunity> and as defined in three different resources: MeSh - Medical Subject Headings 5 , Computer Retrieval of Information on Scientific Projects (CRISP) Thesaurus 6 , and the Webster Medical dictionary 7 . Let us start with Medical Subject Headings (MeSH), the controlled vocabulary thesaurus developed by the U.S. National Library of Medicine to ensure unifor‐ mity and consistency in the indexing and cataloguing of biomedical literature. The information provided in MeSH does not constitute a terminology file in the strict sense. Instead, as shown in Figure 3, it provides a scientific explanation rather than a formal definition. Terminological and Specialised Lexicographic Definitions 55 <?page no="56"?> Fig. 3: Entry for brain-gut axis - MeSH Although scientifically accurate, the passage on the brain-gut axis does not constitute an intensional definition, which requires a general category (genus) and essential distinguishing characteristics (differentiae). Instead, it functions as a description: it lacks a genus term, opening with a descriptive phrase “an in‐ teractive network between the gastrointestinal tract and the brain” that explains its function rather than its nature, and includes examples and operational details (e.g., corticotropin-releasing factor, urocortins, irritable bowel syndrome, stress responses) that illustrate rather than define its core conceptual boundaries. Finally, the inclusion of functional implications, such as the mention of dysregulation in disorders, adds contextual value but further distances the text from the formal role of a definition, which should be limited to the essential characteristics of the concept itself. The role of MeSH is to ensure uniform and consistent content tagging for efficient search and retrieval. However, both are considered types of termino‐ logical resources. The description in Figure 3, labelled as a definition and authored by experts for a specialist audience, is intended to explain and elucidate the concept rather than to prescribe its linguistic usage. Therefore, we do not consider it a terminological or specialised lexicographic definition, but rather a detailed terminological description (not a definition) intended to enhance conceptual clarity within a specific field of expertise. Another important medical terminological resource is the Computer Retrie‐ val of Information on Scientific Projects (CRISP) Thesaurus, which was super‐ seded in 2009 by the Research Portfolio Online Reporting Tools (RePORT). The CRISP Thesaurus served as a structured and controlled vocabulary developed to 56 Rute Costa <?page no="57"?> enable the indexing, classification, and retrieval of biomedical research projects funded by the United States Department of Health and Human Services. This resource does not include the term “brain-gut axis”, so instead we searched for the definition of the concept <mucosal immunity>. This definition aligns more closely with the foundational principles of terminological theory and practice. Fig. 4: Entry for mucosal immunity - CRISP This definition in Figure 4 can be regarded as a terminological definition, as it meets the essential characteristics by clearly identifying the genus / immunity/ and specifying the differentia, namely that this immunity is / induced by the mucosal IgA response/ and / provides specific resistance/ / directed against pa‐ thogens/ / targeting particular mucosal sites/ . Moreover, the definition is concise and focused, highlighting the essential characteristics that distinguish this type of immunity from other forms. It avoids explanations or examples. As such, it serves well as a terminological definition within immunology and related biomedical fields. Finally, we present a specialised lexicographic definition of <Alzheimer’s disease> from the Webster’s Medical Dictionary and compare it with the corresponding entry in the CRISP Thesaurus. Figure 5 presents a traditional dictionary entry for the term “Alzheimer’s disease”, offering a detailed and extensive explanatory description. Terminological and Specialised Lexicographic Definitions 57 <?page no="58"?> Fig. 5: Entry for Alzheimer’s disease - Webster Dictionary In contrast, Figure 6 provides a more concise and linguistically precise formu‐ lation, with reduced interpretative content, aligning more closely with the conventions of a terminological definition. Fig. 6: Entry for Alzheimer’s disease - CRISP 58 Rute Costa <?page no="59"?> Definitions 1 (Fig. 5) and 2 (Fig. 6) differ significantly in their connotative content. Definition 1 incorporates vagueness and ambiguity in phrases such as unknown cause, most common form of dementia, and changes in personality and mood, which may evoke feelings of uncertainty, social urgency, and personal loss. These connotative elements enrich the description by emphasising the human and social impact of the condition beyond its clinical aspects. In contrast, Definition 2 is more technical. It focuses on precise pathological characteristics such as / neurodegenerative disorder of the CNS/ and / brain lesions/ without any emotional or cultural connotations. Its language is neutral, describing symptoms and onset age factually without eliciting empathy or fear. Overall, while Definition 1 blends scientific detail with social and emotional nuances, Definition 2 offers a clear, concise, and denotative description suited for scientific and terminological contexts. This contrast illustrates how definitions can vary depending on their purpose and audience, whether to inform specialists neutrally or to communicate the broader human experience of a disease. We argue that a well-formulated definition expressed in natural language should be translated into a formal or, at the very least, semi-formal representa‐ tion, and, conversely, that formal definitions should be expressed in natural language. This bidirectional conversion is, in our view, essential for maintaining consistency and coherence within both lexical and concept systems. Building on the preceding discussion, we believe that Definition 2, which more closely resembles a terminological definition, can be further refined. To better align with the ISO 704 (2022) standard, we propose the following revision: Definition: neurodegenerative disorder of the central nervous system resul‐ ting in progressive loss of memory and intellectual functions Note 1: The Alzheimer’s disease begins in the middle or later years; charac‐ terised by brain lesions such as neurofibrillary tangles and neuritic plaques. This structure clearly distinguishes the term, its definition (comprising genus and differentia), and supplementary information, as noted, which is essential in terminological work. Grounding definitions in this structured approach is particularly valuable when working with formal languages such as OWL, as it ensures precise and unambiguous representations of concepts and their interrelationships. Unlike informal or purely natural language definitions, formal definitions enable consistent interpretation across different systems and stakeholders. This precision facilitates interoperability, automated reasoning, and effective knowledge management within and across domains. Terminological and Specialised Lexicographic Definitions 59 <?page no="60"?> 8 Declaration on Generative AI: During the preparation of this work, the author used ChatGPT-4 for grammar and spelling checks. The authors have subsequently reviewed and edited the content and take full responsibility for the publication’s final version. 6 Concluding Remarks In conclusion, this study has examined the key distinctions between specialised lexicographic and terminological definitions, differences that are often overloo‐ ked, even by professionals relying on linguistic and terminological resources. Although interconnected, the two types of definitions differ primarily in purpose. Specialised lexicographic definitions, designed for human users, focus on conveying semantic and contextual information and typically avoid forma‐ lisation. Terminological definitions, by contrast, aim to establish a precise and stable relationship between a term and its concept within a structured concept system. Both support communication, but terminological definitions additionally enable language-independent representation and interoperability across lingu‐ istic and computational systems (ISO 704 2022), extending beyond the scope of specialised lexicography. 8 Bibliography Bloomfield, Leonard (1933). Language. London: George Allen & Unwin. Budin, Gerhard (2007). L’apport de la philosophie autrichienne au développement de la théorie de la terminologie: ontologie, théories de la connaissance et de l’objet. In: Savatovsky, Dan/ Candel, Danielle (Eds.). Genèses de la terminologie contemporaine, 1850-1980 (sources et réception), Langages 168, 11-23. Budin, Gerhard (2022). Wüsters Allgemeine Terminologielehre - ein Grenzgebiet zwi‐ schen Sprachwissenschaft, Logik, Ontologie, Informatik und den Sachwissenschaften. In: Candel, Danielle/ Samain, Didier/ Savatovsky, Dan (Eds.). Eugen Wüster et la terminologie de l’école de Vienne. Paris: SHESL (HEL Livres, 2), 243-294. Cabré, Teresa (1999). Terminology: Theory, Methods and Applications. In: Sager, Juan C. (Ed.). Amsterdam/ Philadelphia: John Benjamins. Costa, Rute (2013). Terminology and specialised lexicography: two complementary domains. Lexicographica, 29, 29-42. Costa, Rute/ Vezzani, Frederica (2024). La connotation et la dénotation en terminologie : présentation. Cahiers de lexicologie, 125(2), 11-17. Cruse, David Alan (2001 [1986]). Lexical Semantics. Cambridge: Cambridge University Press. Dotoli, Giovanni/ Boccuzi, Celeste (2012). Définition et dictionnaire. Paris: Hermann. 60 Rute Costa <?page no="61"?> Felber, Helmut (1987 [1984]). Manuel de terminologie. Paris: UNESCO. Frege, Gottlob (1971 [1892]). Sens et dénotatio. In: Écrits logiques et philosophiques. Paris: Seuil. Gaudin, François (1993). Pour une socioterminologie. Rouen: Publications de l’Université de Rouen. Gary-Prieur, Marie-Noëlle (1971). La notion de connotation(s). Littérature, 4, 96-107 Humbley, John (2017). Specialised dictionaries. In: Pedro A. Fuertes-Oliveira (Ed.). The Routledge Handbook of Lexicography. London/ New York: Routledge & Francis Group, 317-334. ISO 1987 (2019). Terminology work and terminology science - Vocabulary. Geneva: International Organization for Standardization. ISO 704 (2022). Terminology work - Principles and methods. Geneva: International Organization for Standardization. Kerbat-Orecchioni, Catherine (1977). La connotation. Lyon: Presses universitaires de Lyon. Martinet, André (1969). La linguistique. Guide alphabétique. Paris: Denoël. Mel’cuk, Igor/ Polguère, Alain (2016). La définition lexicographique selon la Lexicologie Explicative et Combinatoire. Cahiers de Lexicologie, 2016-2, nº 109, La définition. Paris: Classiques Garnier, 61-92. Molino, Jean (1971). La connotation. La linguistique, 7 (1), 5-30. Mounin, Georges (1963). Les problèmes théoriques de la traduction. Paris: Gallimard. Lauren, Christer/ Myking, Johan/ Picht, Heribert (1998). Terminologie unter der Lupe. Vom Grenzgebiet bis zum Wissenschaftszweig. Vienna: TermNet. Lyons, John (1980 [1978]). Sémantique Linguistique. Paris: Larousse. Nilsson, Henrik (2024). Connotation controversées? Quelques effets de la connotation sur les termes suédois. Cahiers de Lexicologie, 2024-2, nº 125, La connotation de la dénotation en terminologie. Paris: Classiques Garnier, 79-102. Rey, Alain (1979). La terminologie. Noms et notions. Paris: Presses universitaires de France. Roche, Christophe (2012). Should terminology principles be reexamined? In: Aguado de Cea, Guadalupe/ Suárez-Figueroa, Mari Carmen/ García-Castro, Raúl/ Montiel-Pon‐ soda, Elena (Eds.). Proceedings of the 10th Terminology and Knowledge Engineering Conference (TKE 2012): New Frontiers in the Constructive Symbiosis of Terminology and Knowledge Engineering. Madrid, Spain: Universidad Politécnica de Madrid, 17- 32. Santos, Claudia/ Costa, Rute (2015). Domain specificity: semasiological and onomasiolo‐ gical knowledge representation. In: Hugo J. Kockaert / Freddy Steurs (Eds.). Handbook of Terminology, Volume 1. Amsterdam/ Philadelphia: John Benjamins, 153-179. Terminological and Specialised Lexicographic Definitions 61 <?page no="62"?> Savatovsky, Dan (2022). Wüster/ Carnap: Vienna School/ Vienna Circle. Terminology bet‐ ween Linguistics and Philosophy of Language. In: Danielle Candel/ Didier Samain/ Dan Savatovsky (Eds.). Eugen Wüster et la terminologie de l’école de Vienne. 141-272. Paris: SHESL (HEL Livres, 2). Seppäla, Selja (2007). La définition en terminologie: typologies et critères définitoires. Théories et Applications. Actes de la conférence TOTh 2007, Le Bourget du Lac, 1er juin 2007. Annecy: Institut Porphyre, Savoirs et Connaissance, 23-44. Temmerman, Rita (2020). Towards New Ways of Terminology Description: The Socio‐ cognitive Approach. Amsterdam/ Philadelphia: John Benjamins. Wüster, Eugen (1979). Einführung in die Allgemeine Terminologielehre und Terminolo‐ gische Lexikographie (Schriftreihe der Technischen Universität Wien 8-1). Wien/ New York: Springer Verlag 62 Rute Costa <?page no="63"?> 1 This is an abridged and revised version of an earlier paper that the authors wrote together with Gerhard Budin: Löckinger, Georg/ Kockaert, Hendrik J./ Budin, Gerhard (2015). Intensional definitions. In: Kockaert, Hendrik J./ Steurs, Frieda (Eds.). Handbook of Terminology. Volume 1. Amsterdam/ Philadelphia: John Benjamins, 60-81. Intensional definitions and concept modelling 1 Georg Löckinger, Hendrik J. Kockaert Abstract: The present chapter deals with intensional definitions, which are one of the most widely used types of definition. While it is necessary to discuss the relevant theoretical background, the present chapter is mainly meant to be a concise, yet comprehensive practical tool for language professionals and domain experts alike. They often have the task of writing intensional definitions in various text production and terminology management settings. Consequently, the present chapter serves a twofold purpose: (1) to summarize the state-of-the-art knowledge on intensional definitions from a terminology science and terminology management point of view and (2) to explore the relationship between intensional definitions and concept modelling by means of the Unified Modeling Language. Also, future options to semi-automatically exploit intensional definitions are outlined. 1 Introduction In human communication - spoken, written, or signed - designations are used to describe objects and concepts. When these designations are unclear or when background knowledge is lacking, definitions become essential for precise understanding. Although other descriptive tools exist (e.g. contexts, encyclopaedic descriptions), definitions are preferred in terminology manage‐ ment. The present chapter focuses on intensional definitions, which have a long-standing tradition and are useful for consistently portraying concepts using natural language. Section 2 gives an overview of intensional definitions in terminology science and management. Section 3 offers practical examples, <?page no="64"?> 2 ISO/ IEC - 19505-1 2012, ISO/ IEC - 19505-2 2012. and Section 4 highlights the universal applicability of intensional definitions. Section 5 and Section 6 explore the relationship between intensional definitions and concept modelling by means of the Unified Modeling Language (UML 2 ). Section-7 provides a brief summary. 2 Theoretical background To describe the real world and the imagined world, we use concepts. Concepts are “unit[s] of knowledge created by a unique combination of characteristics” (ISO 1087 2019: 3). In other words, concepts are mental images of one or several real or imagined objects. The properties of the object(s) in question are converted to characteristics, which may be defined as “abstraction[s] of a property” (ISO 1087 2019: 3). The relevant characteristics are then combined to form concepts. This process is called conceptualization and constitutes an essential building block both in cognitive science and terminology science. It may be visualized as shown in Figure-1. Fig. 1: The process of conceptualization, based on ISO 704 (2022: 5) 64 Georg Löckinger, Hendrik J. Kockaert <?page no="65"?> Let us illustrate this process using the concept “alarm clock”. We start this process by observing alarm clocks as real-world objects that have different properties (shape, colour, sound, etc.). We select properties that are shared by all the objects (alarm clocks) and that are necessarily present in all these objects. These shared properties are then abstracted in our mind into characteristics that make up a concept “alarm clock”. When it comes to defining concepts, we select those characteristics that are essential to understand the concept, and that allow us to distinguish e.g. “railway clock” from “alarm clock”. As stated in Section 1, we use designations to refer to concepts and/ or objects. Often, however, the concept to be described is unknown or must be characterized in greater detail to facilitate successful communication. This is where the definition comes in as a linguistic tool to clarify which concept is referred to by a given designation. A definition is a “representation of a concept […] by an expression that describes it and differentiates it from related concepts” (ISO 1087 2019: 6). Consequently, a definition makes it possible to fix the boundaries of a given concept in relation to other, similar or different concepts. For this purpose, definitions exist in various forms and exploit different features of concepts. Two essential features are those of extension and intension: While a concept’s extension is the “set of all of the objects […] to which a concept […] corresponds”, a concept’s intension is defined as the “set of characteristics […] that make up a concept” (ISO 1087 2019: 1, 3). 3 Intensional definitions: selected examples In ISO 704 (2022: 2), an intensional definition is defined as a “definition that conveys the intension of a concept by stating the immediate superordinate concept and the delimiting characteristic(s)”. Intensional definitions are usually based on generic relations in which the immediate superordinate concept is called the generic concept and the immediate subordinate concept the specific concept. The intension of the specific concept includes the intension of the generic concept plus at least one additional delimiting characteristic (ISO 704 2022: 37). Delimiting characteristics make it possible to set a given concept apart from other concepts, in particular superordinate concepts and coordinate concepts (ISO 1087 2019: -3). In intensional definitions based on partitive relations, the immediate super‐ ordinate concept is named a comprehensive concept: a comprehensive concept is considered the whole and contains the concept to be defined as a part (ISO 704 2022: 37). Intensional definitions and concept modelling 65 <?page no="66"?> It should be noted that apart from the term intensional definition, other terms are used, such as terminological definition (Pavel and Nolet 2001: 23 f), definition by genus and difference, analytical definition (Copi and Cohen 2009: 105 ff.) or definition by intension (Felber 1984: 160 f.). According to ISO 704 (2022: 35 ff.), an intensional definition based on a generic relation is formulated as in Figure 2. Fig. 2: Formulation of an intensional definition according to ISO 704 (2022: 34 ff.), based on Food and Agriculture Organization of the United Nations (2025) The following examples illustrate how intensional definitions are usually formed. Example 1 is an intensional definition based on a generic relation, quoted from Freeman (1997: 56) and slightly modified: (1) convention treaty, usually between more than two States, concerning matters of mutual interest 1. Immediate superordinate concept (generic concept): “treaty” 2. Delimiting characteristics: a. “signed between more than two States” b. “concerns matters of mutual interest” Example 2 is an intensional definition based on a partitive relation, quoted from Aiken (2002: 421) and slightly modified: 66 Georg Löckinger, Hendrik J. Kockaert <?page no="67"?> 3 “The domain, subject or a special usage can be indicated in angle brackets (<>) at the beginning of the definition” (ISO 704 2022: 34; formatting slightly modified). (2) printhead part of an impact printer that contains the pins or other components that force ink from a ribbon onto paper 1. Immediate superordinate concept (comprehensive concept): “impact prin‐ ter” 2. Delimiting characteristics: a. “contains pins or other components” b. “forces ink from a ribbon onto paper” As can be seen from these examples, intensional definitions based on generic relations have the following pattern: generic concept + delimiting characteris‐ tics (Dubuc 2002: 98; Felber 1984: 161; Pavel and Nolet 2001: 24; Wüster 1991: 34). According to ISO 704 (2022: 37), intensional definitions based on partitive rela‐ tions start “with a wording such as ‘part of ’, ‘component of ’, ‘section of ’, ‘period of ’ or ‘element in’”. They may add delimiting characteristics of the partitive concept in order to “allow the [comprehensive concept] to be distinguished from other similar comprehensive concepts” (ISO 704 2022: 16). The delimiting characteristics must be selected according to the specific needs of terminology management and depending on the perspective that the target group intends to apply to a given concept (ISO 704 2022: 33). Example 4 is quoted from Association of the British Pharmaceutical Industry (2025) and slightly modified 3 : (4) water <chemistry> chemical compound consisting of two hydrogen atoms covalently bonded to one oxygen atom Example 5 is quoted from Conference of Translation Services of European States (2002: 29 f.) and slightly modified: (5) water <physics> liquid with a freezing point of 0°C and a boiling point of 100°C at a pressure of 1 atm Intensional definitions and concept modelling 67 <?page no="68"?> 4 Universal applicability Intensional definitions have been used for centuries (Smith 2022; NdiKimbi 1994), and they have proven to be a useful linguistic tool to describe concepts in natural language. Not only in publications on terminology science and terminology management have intensional definitions acquired the status of a standard way of illustrating concepts. There are many other domains of human activity in which intensional definitions have been used or recommended for use, such as logic (Copi and Cohen 2009: 105 ff.), business modelling (Witt 2011: 246), standardization (DIN 2330 2022: 13; ISO 704 2022: 33; ISO 10241-1 2011: 26; ÖNORM A 2704 2015: 44), concept modelling (ISO 24156-1 2014), concept system modelling (ISO/ IEC 5394 2024) and chemistry (Olesen et al. 1997). Beyond more specific scenarios, intensional definitions can be found in almost any kind of textual reference work such as encyclopaedias, general-language and special-language dictionaries, thesauri, terminology databases, classification systems, etc. Numerous publications in the field of terminology science and termino‐ logy management regard intensional definitions as the most important type of definition (Arntz et al. 2014: 65; DIN 2342 2022: 9; Hohnhold 1990: 49; ISO 704 2022: 33). This is due to the fact that intensional definitions have several advantages over other types of conceptual descriptions. Written according to the rules of ISO 704 (2022), they can be an appropriate linguistic tool for describing concepts for at least one of the following reasons: 1. They contain a reference to the (immediate) superordinate concept, which makes it possible to situate the concept in question within the concept system of the relevant domain (Arntz et al. 2014: 66 f.; ISO 704 2022: 33; Suonuuti 1997: 16). On this basis, graphic representations of concept systems (concept diagrams or concept models) can be developed to provide a more comprehensive picture of the terminology concerned (ISO 704 2022: 9 ff.; ISO 24156-1 2014); 2. They enumerate those delimiting characteristics that enable terminology users to differentiate the concept in question from other, especially coor‐ dinate concepts (ISO 704 2022: 33; ISO 1087 2019: 3, 5); 3. They provide assistance in forming appropriate designations (Arntz et al. 2014: 67); 4. They highlight characteristics that would otherwise be hidden (Mihatsch 2006: 154); 5. They can be used effectively to establish the degree of correspondence between concepts, or, in other words, the degree of synonymy or equiva‐ 68 Georg Löckinger, Hendrik J. Kockaert <?page no="69"?> 4 Extensional definition: “definition […] that enumerates all the subordinate concepts […] of a superordinate concept […] under one criterion of subdivision” (ISO 1087 2019: 7). lence between designations. A terminological analysis to this end may be necessary both in monolingual and multilingual settings such as transla‐ tion-oriented terminology management or terminology standardization, just to name two possible scenarios. Methodological guidance on this terminology management task is laid down in the relevant literature (Arntz et al. 2014: 150 ff.; Felber 1984: 152 ff.; ISO 860 2007; ISO 10241-2 2012). Intensional definitions also have some shortcomings. For example, in contrast to extensional definitions 4 , intensional definitions do not contain any of the designations that are used to describe subordinate concepts. In other words, they are often more abstract than extensional definitions, which can make it difficult to deduce related concepts from the definition itself (except for the superordinate concept explicitly referenced). Furthermore, consistent drafting of intensional definitions is necessary for a systematic approach to terminology management. However, by referring to other concepts defined elsewhere, a given intensional definition might necessitate the reading of one or several other intensional definitions to fully grasp the concept concerned. 5 From UML to intensional definitions In UML, the top compartment of the class symbol contains the class name in centred, bold and capitalized formatting. The second compartment lists one or several attribute-value pairs of the class in question (Figure 3a). Class attribute = value attribute = value FishingVessel gear = fishingGear activity = catchesFish The ‘attribute = value’ pairs specify the class in question. The ‘gear = fishingGear’ and the ‘activity = catchesFish’ pairs specify the class ‘Fish- Fig. 3a: Modelling a class in UML notation (ISO 24156-1 2014: 4) ISO 24156-1 converts this UML modelling scheme into a user-defined template. This template represents concepts and characteristics according to the ISO 704 Intensional definitions and concept modelling 69 <?page no="70"?> principles (ISO 24156-1 2014: 4). The top compartment contains the designation referring to the concept in bold and centred formatting, in regular letters. The second compartment lists the attribute-value pairs, considered as representing the essential and/ or delimiting characteristic(s) of the concept in question. An ellipsis indicates that more characteristics might be relevant, depending on the concepts, the criteria of subdivision and the concept system (Figure 3b). concept attribute = value attribute = value … fishing vessel gear = fishing gear activity = catches fish … The ‘attribute = value’ pairs correspond to the characteristics of the concept in question. The ‘gear = fishing gear’ and the ‘activity = catches fish’ pairs correspond to the characteristics of the concept ‘fishing vessel’. Fig. 3b: Modelling a concept in the ISO 24156-1 user-defined UML notation (ISO 24156-1 2014: 4) A generic relation between two concepts is modelled as shown in Figure 4. generic concept specific concept Fig. 4: Modelling a generic relation in the ISO 24156-1 user-defined UML notation using the UML generalization symbol (a solid line ending with an unfilled triangular arrowhead To better visualize the characteristics of concepts and to facilitate the machine processing of UML concept models, characteristics are specified in a more formalized manner by means of one or several attribute-value pairs (Figure 5): 70 Georg Löckinger, Hendrik J. Kockaert <?page no="71"?> Fig. 5: Modelling concepts and generic relations in the ISO 24156-1 user-defined UML notation using the attribute-value specification (ISO 24156-1 2014: 8) To illustrate this, let us take the concept “fishing vessel” as an example: Figure 6 shows that the designation is “fishing vessel” and that this concept consists of two characteristics: (1) “the gear of a fishing vessel is a fishing gear” and (2) “a fishing vessel is used for catching fish”. Intensional definitions and concept modelling 71 <?page no="72"?> 5 For the sake of simplicity, the UML class symbol for “vessel” does not display the concept’s characteristics. Fig. 6: Modelling the concept “fishing vessel” in the ISO 24156-1 user-defined UML notation (ISO 24156-1 2014: 9, slightly modified) In Figure 6, the UML generalization symbol, combined with the UML class symbol for “vessel” 5 , shows us in which way ISO 24156-1 (2014) adopts the UML symbols in a user-defined notation: it informs us that “vessel” is the generic concept of “fishing vessel”, which in turn is the specific concept in this concept system. Given the necessary background knowledge about concept modelling using UML, a language professional or domain expert can easily transform the UML class diagram in Figure 6 into a natural-language definition: fishing vessel vessel which has a fishing gear and is used for catching fish Concept models using UML may be drawn in dedicated software modelling tools. These tools can interpret the semantics behind the various elements of a concept model. Consequently, it should be possible to semi-automatically transform UML concept models into natural-language definitions. This principle also applies to concept models consisting of generic relations, which could be 72 Georg Löckinger, Hendrik J. Kockaert <?page no="73"?> transformed into intensional definitions (or at least into parts thereof). It is a current research desideratum to examine closely the semi-automatic conversion of UML concept models into intensional definitions, both for language profes‐ sionals (terminologists, technical communicators, etc.) and experts in various domains of human activity. 6 From intensional definitions to UML concept models Kop (2008: 157) argues that in controlled English, there are certain language patterns and “keywords [which] allow seeing a sentence like a template with slots”. As is widely known, language patterns, in particular definition patterns, also exist outside controlled language, i.e. in natural language in general. Using the above definition of the concept “fishing vessel”, we can illustrate this point: fishing vessel vessel which has a fishing gear and is used for catching fish This definition can be divided into different slots: [vessel] <which> / has a fishing gear/ (and) / is used for catching fish/ . This division can be abstracted into the following pattern: [generic concept designation] <relative pronoun> / characteristic 1/ (and) / characte‐ ristic 2/ . Now, if this pattern information was embedded in electronic documents and if we had appropriate software tools, we could semi-automatically transform the intensional definition of “fishing vessel” into a UML concept model according to ISO 24156-1 (2014). To our knowledge, this has not yet been implemented anywhere, although there are several studies that treat the computer-assisted transformation of natural language into concept relations of one sort or another (Gromann et al. 2023; Omar and Baryannis 2020; Hossain 2020; Thonggoom 2011). Hebenstreit (2013) presents some promising ideas at the interface between corpus linguistics and terminology management. The author proposes annota‐ ting texts to explicitly identify elements of terminological interest, such as terms, term candidates, information on characteristics of concepts, information on intraconceptual relations, information on interconceptual relations as well as term components and collocates. For identifying characteristics, Hebenstreit (2013) suggests the following XML-based annotation pattern: Intensional definitions and concept modelling 73 <?page no="74"?> <char concept="concept of interest">chunk of text</ char>. Based on this pattern, the intensional definition of the concept “fishing vessel” could be annotated as follows: vessel which <char concept="fishing vessel">has a fishing gear</ char> and <char concept="fishing vessel”> is used for catching fish</ char>. To systematically exploit these possibilities in the future, at least three strands of research are needed: 1. Creating a markup language for terminological elements in running text and developing appropriate annotation tools; 2. Creating well-thought out transformation rules to carry out the conversion from annotated running text to UML concept models; 3. Developing/ extending UML modelling tools that can interpret these trans‐ formation rules correctly for semi-automatic conversion. 7 Conclusion Intensional definitions have been used for centuries as a linguistic tool to describe concepts. Over time, a set of rules has been established that can assist language professionals and domain experts alike in writing or assessing intensional definitions. These rules have been summarized and enriched by examples in the present chapter, so as to provide language professionals and domain experts with a reference tool for practical use. Given the universal applicability of intensional definitions, some ideas have been presented as well which point to future research on the semiautomatic transformation of intensional definitions into UML concept models, and vice versa. Bibliography Aiken, Peter (2002). Microsoft Computer Dictionary. Redmond, WA: Microsoft Press. Arntz, Reiner/ Picht, Heribert/ Schmitz, Klaus-Dirk (2014). Einführung in die Terminolo‐ giearbeit. Hildesheim: Olms. Association of the British Pharmaceutical Industry (2025). Chemistry of Life. Water: a Key Molecule for Life. Available at: https: / / www.abpischools.org.uk/ topics/ chemistr y-of-life/ water-a-key-molecule-for-life/ (Accessed: 11.09.2025) Conference of Translation Services of European States (2002). Recommendations for Terminology Work. Bern: Federal Chancellery. Copi, Irving M./ Cohen, Carl (2009). Introduction to Logic. Prentice Hall. 74 Georg Löckinger, Hendrik J. Kockaert <?page no="75"?> DIN 2330 (2022). Terminologiearbeit - Grundsätze und Methoden. Berlin: Beuth. DIN 2342 (2022). Terminologiewissenschaft und Terminologiearbeit - Begriffe. Berlin: Beuth. Dubuc, Robert (2002). Manuel pratique de terminologie. Montréal: Linguatech éditeur. Felber, Helmut (1984). Terminology Manual. Paris: UNESCO. https: / / unesdoc.unesco.org / ark: / 48223/ pf0000062033.locale=enhttps: / / unesdoc.unesco.org/ ark: / 48223/ pf0000062 033.locale=en. Food and Agriculture Organization of the United Nations (2025). Fishing Vessel Types. Beam trawlers. Technology Fact Sheets. Available at: https: / / www.fao.org/ fishery/ en / vesseltype/ 120/ en (Accessed: 15.09.2025) Freeman, Chas W. (1997). The Diplomat’s Dictionary. Washington, DC: United States Institute of Peace Press. Gromann, Dagmar/ Wachowiak, Lennart/ Lang, Christian/ Heinisch, Barbara (2023). Ext‐ racting Terminological Concept Systems from Natural Language Text. In: Rehm, Georg (Ed.). European Language Grid. A Language Technology Platform for Multilin‐ gual Europe. Cham: Springer, 339-354. Hebenstreit, Gernot (2013). Terminological - A Perspective for Terminology Research and Training? In: Budin, Gerhard/ Humbley, John/ Laurén, Christer/ Lušicky, Vesna (Eds.). Languages for Special Purposes in a Multilingual, Transcultural World. Book of Abstracts of the 19th European Symposium on Languages for Special Purposes, Vienna, 8-10 July 2013. Vienna: University of Vienna, 43. Hohnhold, Ingo (1990). Übersetzungsorientierte Terminologiearbeit. Eine Grundlegung für Praktiker. Stuttgart: InTra-Fachübersetzergenossenschaft. Hossain, Bayzid Ashik (2024). Specifying and Verbalising Conceptual Models Using Controlled Natural Language. PhD dissertation. Macquarie University. ISO-10241-1 (2011). Terminological entries in standards - Part 1: General requirements and examples of presentation. Geneva: International Organization for Standardiza‐ tion. ISO- 10241-2 (2012). Terminological entries in standards - Part 2: Adoption of standar‐ dized terminological entries. Geneva: International Organization for Standardization. ISO- 1087 (2019). Terminology work and terminology science - Vocabulary. Geneva: International Organization for Standardization. ISO- 24156-1 (2014). Graphic notations for concept modelling in terminology work - Part 1: Guidelines for using UML notation in terminology work. Geneva: International Organization for Standardization. ISO- 704 (2022). Terminology work - Principles and methods. Geneva: International Organization for Standardization. ISO- 860 (2007). Terminology work - Harmonization of concepts and terms. Geneva: International Organization for Standardization. Intensional definitions and concept modelling 75 <?page no="76"?> ISO/ IEC-19505-1 (2012). Information technology - Object Management Group Unified Modeling Language (OMG UML). - Part 1: Infrastructure. Geneva: International Organization for Standardization. ISO/ IEC-19505-2 (2012). Information technology - Object Management Group Unified Modeling Language (OMG UML). - Part 2: Superstructure. Geneva: International Organization for Standardization. ISO/ IEC- 5394 (2024). Information technology - Criteria for concept systems. Geneva: International Organization for Standardization. Kop, Christian (2008). Conceptual Modelling Tool for Novice Designers. International Journal of Metadata, Semantics and Ontologies, 3(2), 151-165. Mihatsch, Wiltrud (2006). Kognitive Grundlagen lexikalischer Hierarchien. Untersucht am Beispiel des Französischen und Spanischen. Tübingen: Niemeyer. Ndi-Kimbi, Augustin (1994). Guidelines for Terminological Definitions: The Adherence to and Deviation from Existing Rules in BS/ ISO 2382: Data Processing and Informa‐ tion Technology Vocabulary. Terminology. International Journal of Theoretical and Applied Issues in Specialized Communication, 1(2), 327-350. Olesen, Henrik/ Desmond, Kenny/ Dykbær, René/ Ibsen, Inge/ Bruunshuus, Ivan/ Fuen‐ tes-Arderiu, Xavier/ Hill, Gilbert/ de Araujo, Pedro Soares/ McDonald, Clem (1997). Pro‐ perties and Units in the Clinical Laboratory Sciences. Part XI. Coding Systems - Struc‐ ture and Guidelines. Pure and Applied Chemistry, 69(12), 2607-2620. Omar, Mussa/ Baryannis, George (2020). Semi-automated Development of Conceptual Models from Natural Language Text. Data & Knowledge Engineering, 127, 101796. ÖNORM A 2704 (2015). Terminologiearbeit - Grundsätze und Methoden. Wien: Austrian Standards Institute. Pavel, Silvia/ Nolet, Diane (2001). Handbook of Terminology. Hull: Public Works and Government Services Canada. Smith, Robin (2022). Aristotle’s Logic. In: Zalta, Edward N./ Nodelman, Uri (Eds.). The Stanford Encyclopedia of Philosophy. Stanford University. Suonuuti, Heidi (1997). Guide to Terminology. Nordterm 8. Helsinki: Tekniikan Sanasto‐ keskus. Thonggoom, Ornsiri (2011). Semi-automatic Conceptual Data Modeling Using Entity and Relationship Instance Repositories. PhD dissertation. Drexel University. Witt, Graham C. (2011). Writing Effective Business Rules. A Practical Method. Waltham, MA: Morgan Kaufmann. Wüster, Eugen (1991). Einführung in die allgemeine Terminologielehre und terminolo‐ gische Lexikographie. 3rd ed. Tübingen: Romanistischer Verlag. 76 Georg Löckinger, Hendrik J. Kockaert <?page no="77"?> Research Infrastructures and Terminology: A Sustainability Perspective Tanja Wissik Die Grenzen meiner Sprache bedeuten die Grenzen meiner Welt. L U D W I G W I T T G E N S T E I N (1889-1951) Abstract: This contribution examines sustainability in terminology work through the lens of research infrastructures (RIs). It first discusses the different dimensions of sustainability in terminology practice: the data dimension, the technical dimension, the knowledge dimension, the user dimension, and the organisational dimension. It then analyses how existing RI components and initiatives can support each of these dimensions. Focusing on the European research infrastructures CLARIN and DARIAH, as well as the Austrian consortium CLARIAH-AT, the study demonstrates how these infrastructures contribute to sustainable terminology practices through repositories, standards support, knowledge-sharing infrastruc‐ tures, training initiatives, and funding schemes. The article concludes that collaboration between terminology stakeholders and RIs could be strengthened in order to facilitate the long-term accessibility, usability, and impact of terminology resources and services. 1 Introduction Terminology work increasingly faces challenges related to the sustainability of digital resources and practices. Parallelly, Research Infrastructures (RIs) such as CLARIN (Common Language Resources and Technology Infrastructure) and DARIAH (Digital Research Infrastructure for the Arts and Humanities) have emerged to support digital research in the Social Sciences and Humanities (SSH) by offering both technical and social infrastructures. Despite their potential synergy, there is little research on how RIs can address sustainability challenges <?page no="78"?> in terminology work. This paper integrates two perspectives: the dimensions of sustainability in terminology practice and the infrastructural support provided by research infrastructures. It aims to investigate how RIs can contribute to sustainable terminology practices. The article is structured as follows: After an overview of sustainability in terminology work and the description of the different dimensions of sustaina‐ bility, research infrastructures will be introduced and it will be discussed how RIs can support sustainability across the dimensions identified. The article will end with some concluding remarks. 2 Sustainability in terminology work Sustainability plays a critical role in terminological practice, as digital termino‐ logical resources are typically intended for longterm availability to facilitate potential (re)use by various user groups. The preservation of information and knowledge is not a novel endeavour: institutions such as libraries and archives have engaged in this task for millennia. But, the digital format of terminological resources, as also in other digital-enabled disciplines, introduces specific sustainability challenges: “Nowadays new data are ‛digitally born’, but it is not less complex to guarantee interoperability and sustainability for these new data. Today we pursue an integrated approach to all data, under the perspective of IT & media convergence, as well as from the point of view of workflows […]” (Budin 2015). However, the literature addressing sustainability in terminology practice remains scarce (Wissik 2024). Although the term sustainability is increasingly used in academic discourse, there are various definitions. In the context of research data, sustainability is often understood as data to remain accessible and usable across both technological changes and over time by future generations ( Jarvenpaa and Essén 2023). In the domain of digital humanities, sustainability can be defined as project outcomes “being optimsed for reuse” (Edmond and Morselli 2020). This is also applicable to terminology projects, since they often contain „konkrete Ergebnisse, die in der zuvor untersuchten Praxis auch zum Einsatz kommen, sei es in Form eines Fachglossars, einer Terminologiedatenbank, einer Software für die Verarbeitung terminologischer Daten oder für die Extraktion solcher Daten aus Texten bzw. aus Korpora” (Budin 2019: 8). Furthermore, the multidimensional nature of sustainability is a recurring theme in the literature (Lavoie 2012; Neuroth and Rapp 2016; Edmond and Morselli 2020; Branco et al. 2023; Wissik 2024). Besides sustainability, also terms like maintenance are used in standards (e.g., ISO 26162-2: 2019; ISO 26162-3: 2023; ISO 12616-1: 2021) and by several authors (e.g. Pilke et al. 78 Tanja Wissik <?page no="79"?> 1 https: / / www.w3.org/ TR/ skos-reference/ 2 https: / / www.w3.org/ 2016/ 05/ ontolex/ 2021; Chiocchetti et al. 2023). As already mentioned, sustainability is multidi‐ mensional. In this section the dimensions of sustainability in terminology as described by Wissik (2024), will be discussed: the data dimension, the technical dimension, the knowledge dimension, the user dimension and the organisational dimension. 2.1 The data dimension The data dimension deals with both data and metadata. In the context of terminology work, data can refer to an individual terminological record within a termbase, an entire termbase, or an export from a termbase in formats like TBX. Metadata, by contrast, describes the attributes of a terminological entry or the entire resource - such as author, licence, size, languages, format, creation date, or revision date - and plays a crucial role in enabling the retrieval, sharing, and reuse of the resource (Wissik 2024). The data dimension of sustainability is concerned with the preservation of data and metadata. However, preservation is not for its own sake, but as precondition for its reuse. In this respect the FAIR data principles (Wilkinson et al. 2016) play a major role. Vezzani (2022) and Vezzani and di Nunzio (2023) discuss specifically the application of the FAIR data principles to terminological data. Established data and metadata standards are essential for the preservation and reuse of data (Edmond and Morselli 2020; Mattern 2022; Vezzani 2022; Chiochetti et al. 2023: 471). Already Budin and Melby (2000) discussed the issues of accessibility and interoperability of terminological resources. Furthermore, open formats (Holmes and Takeda 2023) should be preferred over proprietary formats. Standards for representing termi‐ nological data can be de facto standards such as Text Encoding Initiative (TEI) or standards published by standards organisations like ISO and their national counterparts. The TEI guidelines include a whole chapter dedicated to encoding dictionaries and there is the TEI Lex-0, a baseline encoding and a target format to facilitate the interoperability of heterogeneously encoded lexical resources. Furthermore, there are the de facto standards from the W3C, such as Simple Knowledge Organization System (SKOS) 1 or OntoLex-Lemon model 2 . Moreover, there are standards by the International Organization for Standardization (ISO), especially the standards created in the Technical Committee ISO/ TC 37, like TermBase eXchange (TBX) for representing and exchanging information on terminological data (ISO 30042 2019). Research Infrastructures and Terminology: A Sustainability Perspective 79 <?page no="80"?> Besides standards, data repositories are needed for data preservation and reuse. On one hand, terminological resources can be archived for the longterm in data repositories and on the other hand, these repositories can function as an additional dissemination channel that supports findability through metadata and citability through persistent identifiers (Wissik 2024). Data preservation is only a snapshot in time, usually the final resource is archived. However, the continuous curation of terminological data is as important as the preservation to ensure that the data remains useful for the intended target group. This includes adding new or updating existing entries, cleaning the data or changing the structure of the resource. It is important that any changes made to the resource are reflected in the metadata in order to ensure transparency and reproducibility (Wissik 2024). 2.2 The technical dimension Several software solutions are supporting terminology work, ranging from terminology management systems over term extraction tools, to computer aided translation systems (Steurs et al. 2015) and lately also generative artificial intelligence (Reineke 2023; San Martín 2024). Regular maintenance ensures that the software can continue to serve its purpose effectively, securely, and efficiently. One aspect of technical maintenance and sustainability is regular software updates so that the software remains functional. Another aspect is the free availability of source code. Open-source software allows complete access to the code of the software, facilitating the maintenance. Besides being accessible, the code should also be understandable by others than the original developers to allow new contributors to extend and update the software. Experiences with legacy code in the past have led to the “Clean Code” movement, developing practices and principles that ease the writing of good code (Latte et al. 2019). However, open-source software among translation technologies and termino‐ logy tools are still rare (Sandrini 2019; Wissik 2024). Related aspects to openness of software and good code readability are also the use of code repositories such as GitHub, and the provision of appropriate technical documentation. Furthermore, more users of a software or service often led to increased error reporting or feature requests, resulting in regular updates. This increases sustainability too (Broeder and Odijk 2022). 80 Tanja Wissik <?page no="81"?> 2.3 The knowledge dimension The knowledge dimension includes the dissemination of knowledge through the documentation of processes, the publication of guidelines, and contributions to the scientific literature (Wissik 2024; Edmond and Morselli 2020; de Jong et al. 2022; Neuroth and Rapp 2016). In the context of terminology work, knowledge encompasses the methodologies employed as well as general procedures and workflows applied in practice. For resources and tools to be effectively utilized, potential users must be informed about their underlying principles and use cases (Branco et al. 2023). This knowledge can be conveyed through scientific articles and presentations, guidelines, manuals, checklists, and similar documentation, as well as through information about the creation and development of specific terminological resources (Wissik 2024). Published responses to user queries regarding terminological problems also contribute to knowledge sharing. These responses, which often provide richer contextual information, may not conform to the structure of a terminology database or dictionary entry and are therefore often disseminated in alternative formats, e.g., as unstructured text on a website (Wissik 2024). 2.4 The user dimension In general, the user dimension includes community building, community enga‐ gement and outreach activities (Wissik 2024; Draxler et al. 2022; Edmond and Morselli 2020; Neuroth and Rapp 2016). In a terminology context, Bhreathnach (2011: 161) states that “[i]nformation about term resources needs to be made available as widely as possible. Word‐of‐mouth alone, though useful, will not spread the information beyond a certain core group. Potential users need to be made aware of term resources and term planning”. Several factors influence this dimension, including dissemination activities as well as training and consulting activities and the creation of training materials. Additionally, the establishment of partnerships and participation in networks play an important role, as summarised in Wissik (2024). 2.5 The organisational dimension The organisational dimension, sometimes also called economic dimensions, encompasses organisational frameworks, organisational continuity and fund‐ ing mechanisms that support the sustainability of resources and services and staff retention (Wissik 2024; Edmond and Morselli 2020; Lavoie 2012). Challenges such as understaffing (Chiocchetti 2021: 119), not enough funding Research Infrastructures and Terminology: A Sustainability Perspective 81 <?page no="82"?> 3 https: / / www.dariah.eu/ network/ members-and-partners/ 4 NCF Meeting minutes from April 19, 2012 () 5 The CLARIAH-AT website is available at https: / / clariah.at and significant differences in funding and project duration depending on the organisational context (e.g., public institution, university or private company) are frequently reported (Drame 2015: 518). 3 Research Infrastructures In the Social Sciences and Humanities (SSH), the European Strategy Forum on Research Infrastructures (ESFRI) identifies five major European Research Infrastructures (RIs) and six Research Infrastructure Projects (ESFRI 2021). This paper focuses on CLARIN and DARIAH as the two RIs are most relevant to terminology research and practice on a European level, and on CLARIAHAT, the national consortium for CLARIN and DARIAH in Austria as a national example. CLARIN, established as an ERIC, a European Research Infrastructure Consortium in 2012, aims to provide a sustainable digital infrastructure for sharing and accessing language data across Europe to support research in the humanities, social sciences, and beyond (de Jong et al. 2022). Currently, CLARIN includes 26 member countries that operate a distributed network of data, service, and knowledge centres. Through this network, CLARIN offers digital language resources, services, and expertise to researchers, scholars, and students across disciplines. Similarly, DARIAH, founded as an ERIC in 2014, seeks to empower research communities through digital methods that facilitate the creation, connection, and sharing of knowledge about culture and society (DARIAH 2025). DARIAH currently comprises 23 member countries and several Cooperating Partners in ten non-member countries. 3 Austria was a founding member of CLARIN and DARIAH. Gerhard Budin played a leading role in this context, taking part in the preparatory phase project of CLARIN and coordinating CLARIN-AT (2010-2013), the Austrian project contributing to CLARIN ERIC. He served as the first CLARIN National Coordinator for Austria, participating in the inaugural meeting of the CLARIN National Coordinators’ Forum (NCF) in April 2012 4 , and held this position until 2014. Since 2014 the Austrian initiatives have been active under the umbrella of CLARIAH-AT. However, only in 2019 the informal CLARIAH-AT network became an official consortium acting under CLARIAH-AT 5 (Wissik et al. 2025). In the following, we will look how different infrastructure components and initiatives support the different dimensions of sustainability. Already Budin (2015) investigated the relation between RIs, in particular CLARIN, and 82 Tanja Wissik <?page no="83"?> 6 https: / / standards.clarin.eu/ sis/ index.xq Computational Terminology and Computational Translation Studies. After 10 years and a changing landscape, it is worthwhile to reassess the findings. 3.1 Supporting the data dimension As described before, sustainable terminology work requires the use of standards, and standardised data and metadata formats. If possible, open standards and formats should be preferred. Furthermore, sustainable terminology work needs comprehensive metadata and reliable data repositories. Regarding standards, there are several overlaps between CLARIN and DA‐ RIAH and where those RIs can support the data dimension of sustainability. Within CLARIN, there are a number multiple thematic committees, and one committee, the Standards and Interoperability Committee (SIC), is especially dedicated to standards (Bánski and Hedeland 2022). Its main tasks include collecting, consolidating and preparing recommendations on standards, formats and related best practices, maintaining the set of standards supported by CLARIN and adapting them to new developments and promoting the use of the recommended standards, formats and best practices (CLARIN 2025a). Additionally, the SIC maintains the Standards Information System (SIS) 6 , an overview of recommended standards used within CLARIN centres. In this overview there are also standard formats relevant to terminology such as TBX, SKOS, RDF, XML or the ISO standard regarding terminology management systems (e.g., ISO 26162-1: 2021). Within DARIAH, one of several thematic Working Groups (WGs), the Lexical resources WG, explores, assesses and recommends standard tools and methods for the creation, application and dissemination of borndigital and retrodigitized lexical resources. Regarding reliable data repositories, within CLARIN there is a network of so-called CLARIN B-Centres, which operate certified data repositories. Those data repositories provide persistent identifiers, metadata and information on the licences and terms of use and therefore support the FAIR data principles that make language resources findable, accessible, interoperable and reusable (Lušicky and Wissik 2019). For the Austrian context, in the CLARIAH-AT con‐ sortium, there are two B-Centres that run data repositories, A Resource Centre for the HumanitiEs (ARCHE) and Geisteswissenschaftliches Asset Management System (GAMS) (Wissik et al. 2025). Even though the use of data repositories supports sustainability, it is not a widespread practice among stakeholders in Research Infrastructures and Terminology: A Sustainability Perspective 83 <?page no="84"?> 7 https: / / vocabs.dariah.eu/ en/ institutional terminology work as an interview study by Wissik (2024) showed. In order to change that, research infrastructures such as CLARIN could reach out to the terminology community with awareness raising activities. Regarding metadata, there are also some relevant initiatives like the CMDI metadata schema (Windhouwer and Goosen 2022) and the DARIAH Vocabs services 7 , hosted and maintained by the Austrian Centre for Digital Humanities (ACDH). The Vocabs services contain on one hand a controlled vocabulary repository, and on the other hand an editor in order to collaboratively create and maintain controlled vocabularies (Wissik 2025). The Vocabs repository contains several controlled vocabularies (e.g., ISO language codes, Austrian Fields of Science and Technology Classification (ÖFOS)) which can be used in the context of metadata creation in order to make metadata more interoperable and therefore also more sustainable. 3.2 Supporting the technical dimension Regarding the technical dimension, the maintainability of software and services, software updates (Broeder and Odijk 2022), and opensource availability (Venters et al. 2014) are crucial. Integrating software and services into the network of infrastructure providers can have several benefits in supporting the technical dimension of sustainability. One benefit is that the developers are integrated into a network where they can exchange knowledge, help each other solving problems, supporting also developers who are not integrated into a developer team at their home institution because there might be only one developer. In CLARIN, there is for example the Technical Centre Committee (TCC) that ensures the consistence, the coherence and the stability of infrastructure services through decisions on implementation as well as coordination between the centres and members (CLARIN 2025b). One of its main tasks is to serve as a forum for CLARIN centres to exchange ideas and experiences. Furthermore, there are several dedicated task forces for specific services (e.g., one for the CLARIN Federated Content Search). Another sustainability benefit when services (e.g., online term extraction tool) are incorporated into an infrastructural network of service providers, is the solidarity among the service providers: if one provider stops running a service, another provider in the infrastructure can take over. Furthermore, RIs such as CLARIN and DARIAH are committed to actively promoting openness i.e. open data, open-source code, and open standards which 84 Tanja Wissik <?page no="85"?> 8 https: / / halshs.archives-ouvertes.fr/ DARIAH 9 https: / / openmethods.dariah.eu/ 10 https: / / www.clarin.eu/ k-centre-catalogue supports sustainability. Especially in the area of terminology there is a need to catch up in this regard and in participating in such RIs, a change of attitude and practices in the community can be facilitated. 3.3 Supporting the knowledge dimension The documentation of workflows and guidelines ensures knowledge transfer and continuity. However, often these documentations are only published on the institutional websites or the website of a specific terminological resource. Research infrastructures provide services that facilitate the sustainable sharing of scientific publications, workflow documentations and guidelines. DARIAH operates several infrastructure components, facilitating the knowledge dimen‐ sion: for example, the DARIAH collection on the HAL repository 8 for DA‐ RIAH-affiliated research outputs, or the OpenMethods 9 platform that provides curated content about Digital Humanities Methods and Tools. Furthermore, since 2024, DARIAH is also operating a peer-reviewed journal, Transformations: A DARIAH Journal. These knowledge sharing platforms can be used to share terminological knowledge, methods and guidelines in a stable way besides publishing it on institutional websites. Another relevant infrastructural com‐ ponent is the SSH Open Marketplace (König et al. 2024), a curated discovery portal for tools, services, workflows and training materials maintained by three RIs DARIAH, CLARIN and CESSDA. Of particular interest is the possibility to create workflows in the SSH Open Market place, to contextualise them with relevant tools, example data and publications and share them with the community. Furthermore, these workflows are licenced and are citable. This way, workflows are not only documented, but they can also be shared with a wider audience. Besides the already described infrastructure, also the CLARIN Knowledge Infrastructure with the CLARIN Knowledge Centres (so called K-Centres) needs to be mentioned. The CLARIN Knowledge Infrastructure is a way to make expertise and knowledge available in an organized way. CLARIN K-Centres are a physical or a virtual place, where people (both those using the infrastructure and those who help building or operating it) can get cross-border access to knowledge and expertise in specific areas, through a helpdesk. 10 While there are several CLARINCentres that provide expertise in terminology among other areas, there is also one specific CLARIN K-Centre for Terminology Research Infrastructures and Terminology: A Sustainability Perspective 85 <?page no="86"?> 11 https: / / www.clarin.eu/ k-centres/ trtc 12 https: / / www.clarin.eu/ content/ learning-hub 13 https: / / campus.dariah.eu/ Resources and Translation Corpora 11 run by the University of Vienna and part of CLARIAH-AT. Another CLARIN instrument that facilitates knowledge sharing are the CLARIN Mobility Grants. The grants support collaboration within CLARIN, with new user communities and with other RIs (CLARIN 2025c) and could be used by the terminology community. 3.4 Supporting the user dimension The user dimension includes on one hand training activities and the availability of training resources and on the other hand user engagement strategies and dis‐ semination activities. Both RIs, CLARIN and DARIAH, provide several training activity initiatives and share training resources. In this context, Open Educati‐ onal Resources (OER) have to be mentioned. OERs are “teaching, learning and research materials in any medium, digital or otherwise, that reside in the public domain or have been released under an open license that permits no-cost access, use, adaptation and redistribution by others with no or limited restrictions” (UNESCO 2012). OERs are often shared on dedicated platforms. CLARIN is collecting training resources related to CLARIN on the CLARIN Learning Hub 12 . DARIAH is running DARIAH-Campus 13 , a hosting and discovery platform for learning resources related to digital methods in the arts and humanities. The learning resources on DARIAH-Campus are assigned a persistent identifier and are therefore citable. First, the terminology community can use already existing learning materials for training purposes especially regarding new digital methods and tools. Second, the platforms can be used by the terminology community to share their own training materials in a sustainable way and make them more accessible and citable. Furthermore, regarding dissemination activities, the different dissemination channels of CLARIN, DARIAH and their national consortia as well as their networks can help to disseminate terminolo‐ gical resources, tools and services and reach out to a wider community. 3.5 Supporting the organisational dimension Stable funding, institutional support, and long term strategies are necessary to sustain terminological resources and services. If terminology resources, tools and services are integrated into a research infrastructure, it is more likely that there is a long-term commitment, because the memberships in research 86 Tanja Wissik <?page no="87"?> 14 https: / / www.dariah.eu/ activities/ dariah-theme/ infrastructures are usually planned for the medium or long term. Furthermore, research infrastructures such as CLARIN and DARIAH provide also small-scale funding opportunities that can support the organisational dimension of sustai‐ nability. One example is the DARIAH Theme funding scheme, i.e., biannual thematic funding calls 14 . Furthermore, also at national level there are funding opportunities: CLARIAH-AT has calls for project funding and for early career funding (Wissik et al. 2025). 4 Concluding remarks Integrating sustainability principles into terminology work is essential in the long term. In this contribution the different dimensions of sustainability were described and it was discussed how research infrastructures such as CLARIN and DARIAH and their national consortia can offer critical support regarding sustainability. Strategic collaborations and awareness raising can bridge the gap between terminology stakeholders and RIs to fully benefit from all the infrastructure components at hand and to ensure that terminology resources, tools and services remain accessible, reusable, and impactful into the future. Bibliography Bański, Piotr/ Hedeland, Hanna (2022). Standards in CLARIN. In: Fišer, Darja/ Witt, Andreas (Eds.). CLARIN: The Infrastructure for Language Resources. Berlin: De Gruyter, 309-339. Bhreathnach, Úna (2011). A best-practice model for term planning. PhD thesis. Dublin City University. Branco, António et al. (2023). The CLARIN infrastructure as an interoperable language technology platform for SSH and beyond. Language Resources and Evaluation, 57, 1-32. Broeder, Daan/ Odijk, Jan (2022). Sustainability and Genericity of CLARIN Services in the Netherlands. In: Fišer, Darja/ Witt, Andreas (Eds.). CLARIN: The Infrastructure for Language Resources. Berlin: De Gruyter, 133-162. Budin, Gerhard (2015). Digital Humanities, Language Industry, and Multilingualism: Global Networking and Innovation in Collaborative Methods. In: Forstner, Mar‐ tin/ Lee-Jahnke, Hannelore (Eds.). CIUTI-Forum 2014: Pooling Academic Excellence with Entrepreneurship for New Partnerships. Lausanne: Peter Lang, 423-448. Research Infrastructures and Terminology: A Sustainability Perspective 87 <?page no="88"?> Budin, Gerhard (2019). Zum Entwicklungsstand der Terminologiewissenschaft. In: Drewer, Petra/ Pulitano, Donatella (Eds.). Terminologie: Epochen, Schwerpunkte, Umsetzungen. Berlin: Springer, 7-19. Budin, Gerhard/ Melby, Alan (2000). Accessibility of Multilingual Terminological Re‐ sources - Current Problems and Prospects for the Future. In: Proceedings of the Second International Conference on Language Resources and Evaluation (LREC 2000). Athens: European Language Resources Association, 1215-1220. Chiocchetti, Elena (2021). Effects of social evolution on terminology policy in South Tyrol. Terminology, 27(1), 110-139. Chiocchetti, Elena/ Lušicky, Vesna/ Wissik, Tanja (2023). Multilingual Legal Terminology Databases: Workflows and Roles. In: Biel, Łucja/ Kockaert, Hendrik J. (Eds.). Handbook of Terminology, vol. 3: Legal Terminology. Amsterdam/ Philadelphia: John Benjamins, 458-484. CLARIN (2025a). CLARIN Standards and Interoperability Committee. Available at: http s: / / www.clarin.eu/ governance/ standards-committee (Accessed: 20.03.2026) CLARIN (2025b). Technical Centres Committee. Available at: https: / / www.clarin.eu/ gov ernance/ technical-centres-committee (Accessed: 20.03.2026) CLARIN (2025c). CLARIN Mobility Grants. Available at: https: / / www.clarin.eu/ content/ clarin-mobility-grants (Accessed: 20.03.2026) DARIAH (2025). Mission & Vision. Available at: https: / / www.dariah.eu/ about/ mission-v ision/ (Accessed: 20.03.2026) De Jong, Franciska/ Van Uytvanck, Dieter/ Frontini, Francesca/ van den Boch, Antal/ Fiser, Darja/ Witt, Andreas (2022). Language Matters. In: Fišer, Darja/ Witt, Andreas (Eds.). CLARIN: The Infrastructure for Language Resources. Berlin: De Gruyter, 31-58. Drame, Anja (2015). The social and organizational context of terminology work. In: Kockaert, Hendrik/ Steurs, Frieda (Eds.). Handbook of Terminology, vol. 1. Amster‐ dam/ Philadelphia: John Benjamins, 507-519. Draxler, Christoph/ Geyken, Alexander/ Hinrichs, Erhard/ Klosa-Kückelhaus, An‐ nette/ Teich, Elke/ Trippel, Thorsten (2022). How to Connect Language Resources, Infrastructures, and Communities. In: Fišer, Darja/ Witt, Andreas (Eds.). CLARIN: The Infrastructure for Language Resources. Berlin: De Gruyter, 275-306. Edmond, Jennifer/ Morselli, Francesca (2020). Sustainability of digital humanities projects as a publication and documentation challenge. Journal of Documentation, 76(5), 1019-1031. ESFRI (2021). Roadmap 2021. Strategy Report on Research Infrastructures. Milano. Avai‐ lable at: https: / / roadmap2021.esfri.eu/ media/ 1295/ esfri-roadmap-2021.pdf (Accessed: 20.03.2026) Holmes, Martin/ Takeda, Joey (2023). From Tamagotchis to Pet Rocks: On Learning to Love Simplicity through the Endings Principles. Digital Humanities Quarterly, 17(1). 88 Tanja Wissik <?page no="89"?> ISO 12616-1 (2021). Terminology work in support of multilingual communication. Part 1: Fundamentals of translation-oriented terminography. Geneva: International Standards Organization. ISO 26162-2 (2019). Management of terminology resources. Terminology databases. Part 2: Software. Geneva: International Standards Organization. ISO 26162-3 (2023). Management of terminology resources. Terminology databases. Part 3: Content. Geneva: International Standards Organization. ISO 30042 (2019). Management of terminology resources - TermBaseEXchange (TBX). Geneva: International Standards Organization. Jarvenpaa, Sirkka L./ Essén, Anna (2023). Data sustainability: Data governance in data infrastructures across technological and human generations. Information and Orga‐ nization, 33(1). König, Alexander/ Barbot, Laure/ Grisot, Cristina/ Kurzmeier, Michael/ Gray, Edward J. (2024). The SSH Open Marketplace and CLARIN. In: Lindén, Krister/ Friser, Darjia (Eds.). Selected Papers of the CLARIN Annual Conference 2023. Linköping: Linköping University Electronic Press, 114-125. Latte, Björn/ Soren, Henning/ Wojcieszak, Maik (2019). Clean Code: On the Use of practices and Tools to Produce Maintainable Code for Long-Living Software. In: EMLS 2019: 6th Collaborative Workshop on Evolution and Maintenance of Long-Living Systems (SE19). Stuttgart: ACM. Lavoie, Brian (2012). Sustainable Research Data. In: Pryor, G. (Ed.). Managing Research Data. London: Facet Publishing, 67-82. Lušicky, Vesna/ Wissik, Tanja (2019). Language resources in repositories and catalogues: Pilot study on their potential for LSP teaching and training. In: Simonnæs, Ingrid et al. (Eds.). New Challenges for Research on Language for Special Purposes. Selected Proceedings from the 21st LSP-Conference, 28-30 June 2017, Bergen, Norway. Berlin: Frank & Timme, 327-342. Mattern, Eleanor (2022). The Linguistic Data Life Cycle, Sustainability of Data, and Principles of Solid Data Management. In: Berez-Kroeker, Andrea L./ McDonnell, Brían/ Koller, Emily/ Collister, Lauren B. (Eds.). The Open Handbook of Linguistic Data Management. Cambridge, MA: MIT Press, 61-72. Neuroth, Heike/ Rapp, Andrea (2016). Nachhaltigkeit von digitalen Forschungsinfra‐ strukturen. Bibliothek - Forschung und Praxis, 40(2), 264-270. Pilke, Nina/ Nissilä, Nina/ Landqvist, Hans (2021). Terminology as a societal resource. Possibilities and responsibilities in a changing world. Terminology, 27(1), 3-9. Reineke, Detlef (2023). Terminologiearbeit mit ChatGPT & Co. tekOm Fachzeitschrift für Terminologie, 19(1), 25-28. Research Infrastructures and Terminology: A Sustainability Perspective 89 <?page no="90"?> Sandrini, Peter (2019). Translationspolitik für Regional- oder Minderheitensprachen. Unter besonderer Berücksichtigung einer Strategie der Offenheit. Berlin: Frank & Timme. San Martín, Antonio (2024). What Generative Artificial Intelligence Means for Termino‐ logical Definitions. In: Proceedings of the 3rd International Conference on Multilin‐ gual Digital Terminology Today (MDTT 2024). Granada: Universidad de Granada. Steurs, Frieda/ De Wachter, Ken/ De Malsche, Evi (2015). Terminology Tools. In: Kockaert, Hendrik J./ Steurs, Frieda (Eds.). Handbook of Terminology, vol. 1. 2nd ed. Amster‐ dam/ Philadelphia: John Benjamins, 222-249. UNESCO (2012). OER Declaration. World Open Educational Resources Congress. Paris: UNESCO. Venters, Colin/ Jay, Caroline/ Lau, Lydia/ Griffiths, Michael K./ Holmes, Violeta/ Ward, Rupert/ Austin, Jim/ Dibsdale, Charlie E./ Xu, Jie (2014). Software Sustainability: The Modern Tower of Babel. In: Proceedings of Third International Workshop on Requi‐ rements Engineering for Sustainable Systems. Karlskrona: ACM, 7-12. Vezzani, Federica (2022). Terminologie numérique: conception, représentation et gestion. Lausanne: Presses polytechniques et universitaires romandes. Vezzani, Federica/ Di Nunzio, Giorgio Maria (2023). Multilingual digital terminology: Introduction to the special issue. Digital Scholarship in the Humanities, 38(1), i1-i5. Wilkinson, Mark D./ Dumontier, Michel/ Aalbersberg, IJsbrand Jan/ Appleton, Gabri‐ elle/ Axton, Myles/ Baak, Arie/ Blomberg, Niklas/ Boiten, Jan-Willem/ Bonino da Silva Santos, Luiz/ Bourne, Philip E./ Bouwman, Jildau/ Brookes, Anthony J./ Clark, Tim/ Crosas, Mercè/ Dillo, Ingrid/ Dumon, Olivier (2016). The FAIR Guiding Principles for scientific data management and stewardship. Scientific Data, 3, 160018. Windhouwer, Menzo/ Goosen, Twan (2022). Component Metadata Infrastructure. In: Fi‐ šer, Darja/ Witt, Andreas (Eds.). CLARIN: The Infrastructure for Language Resources. Berlin: De Gruyter, 191-222. Wissik, Tanja (2024). Dimensions of sustainability in terminology practice in institutional settings. Terminology Science & Research / Terminologie: Science et Recherche, 27, 93-116. Wissik, Tanja (2025). An Infrastructural Approach to Terminology Work: The Case of Research Infrastructures. In: Vandeghinste, Vincent/ Kontino, Thalassia (Eds.). Selected Papers of the CLARIN Annual Conference 2024. Linköping: Linköping University Electronic Press, 1-12. Wissik, Tanja/ Scholger, Walter/ Klenke, Kerstin/ Lušicky, Vesna/ Ďurčo, Matej/ Trognitz, Martina/ Štuhec, Seta/ Steiner, Elisabeth (2025). CLARIAH-AT: Back (and) to the Future. In: Proceedings of the CLARIN Annual Conference 2025, 63-67. 90 Tanja Wissik <?page no="91"?> Terminology work, principles, and applications: Decades of development from Colombia Juan Carlos Diaz Vásquez, María Cecilia Plested Alvarez Wer fremde Sprachen nicht kennt, weiss nichts von seiner eigenen. J O H A N N W O L F G A N G G O E T H E (1749-1832) Abstract: Since 1995, with the constitution of the GITT (Terminology and Translation Research Group), terminological research and teaching activi‐ ties in Colombia, are based on the General Theory of Terminology - GTT (Wüster, 1985; Budin, 1996) supported by academics in Colombia, Austria, Germany, and USA. We created developments in terminology science, work, standardization, translation and interpretation of multilingual spe‐ cialties. The evidence collected during these 30 years crosses a wide variety of linguistic, and field boundaries of knowledge, confirming the current importance of GTT in science by applying its principles and methods. We dive into the basics of Budin’s WIKO model. Its focus can be seen in almost all projects and products related to terminology management at the EAFIT University and GITT from the University of Antioquia, Colombia, over these decades. This is a constant element throughout the developments of terminology in Colombia and shown in this contribution. 1 Introduction The influence of Budin’s work (e.g. 1996, 2000, 2019) has been advancing in the last four decades in Colombia and has deeply impacted many subject fields in science in many ways. First, he has published many journal articles and books in collaboration with other scholars; he has also participated in countless academic events and has had leading roles in educational institutions in Vienna. Second, as tutor and director of doctoral theses, he is restless in his research projects, as <?page no="92"?> in many other research activities. Third, his active participation in international networks, mainly with the technical committee ISO/ TC 37, and his dedication to traveling to many places established long-lasting cooperation networks around the world. His contributions in advancing terminology theory and applications (Budin 2001, 2019; Wright and Budin 1997) are of paramount importance in the field, not only because Budin promoted the basis of modern terminology found in Wüster (1979), but also because he expanded it and intergrated it to relevant fields, such as standardization, terminology and translation (Budin 1996), terminology and digitalization, as well as language, communication and information technologies (Budin 2000, 2015), terminology in multilingual settings (Budin 2002, 2013), terminology in knowledge representation, and organization (Budin 1996, 2006). By doing so, Budin has developed the meaning of terminology that is used in this contribution. The later field of influence is probably the most relevant aspect present at most of the research projects developed under his influence. That is why, we briefly describe the WIKO model as Budin proposed. 2 Early years of terminology work with the ISO/ TC37 standards. As mentioned, Budin’s involvement in international networks contributed to developing terminology and terminology work in Colombia since 1995. Institu‐ tions such as the Copenhagen Business School, the International Information Centre for Terminology (InfoTerm), the International Institute for Terminology Research (IIFT), the Society for Terminology and Knowledge Transfer (GTW), the Kent State University, the University of Applied Sciences of Cologne, the University of Vienna, the University of Antioquia are some of institutions, where his influence has led to further developments. Most of the colleagues who have supported us in our work in Colombia are also members of the most important international organization for standards, the Technical Committee ISO/ TC 37. The technical committee at ISO is in charge of issuing terminology standards. Inspired by Budin, the terminology work here and the international networ‐ king started to take shape through the combination of efforts in different sectors. The academic work done in different Colombian universities, in cooperation with international institutions, and a wide range of applied research projects evidence the impact of terminology work. One of them is introduced in the following. The GITT and the Colombian Terminology Network (ColTerm) have made the aspect of terminology work relevant in the country achieving that many 92 Juan Carlos Diaz Vásquez, María Cecilia Plested Alvarez <?page no="93"?> researchers new to the topic of terminology were trained in conceptual harmo‐ nization, terminology standards, and computer-assisted terminology work. The terminology research work carried out within various universities, and the ideas from Budin regarding terminology management (Budin, 2000) were central to adopting international standards regarding terminology data management. These, along with the wide range of applied research projects, paved the way to establish a methodology to create and maintain terminology databases. This methodology is founded upon ISO/ TC37 standards, such as ISO704 (2022), ISO 860 (2007), ISO 29383 (2002), among others. Theoretically, the terminology work applied by these research groups is based on the WIKO model (Budin 1996), which considers terminology as the core element organizing knowledge, information, and communication systems. This perspective allowed us to collect and revise all the relevant information collecting concepts and definitions attached to terms in specialized texts from a specific domain. This method uses lexicological and terminological analysis to recognize and understand a given specific domain to decompose its object of knowledge from its contextualized terminological components (Plested and Giraldo 2000). Along with this approach, it becomes necessary to mention some of the results at the terminology level, standardization, and this is the Colombian Institute for Standardization and Certification, Icontec. Icontec is a private, nonprofit institute that promotes standardization, met‐ rology, and certification processes in Colombia. This institute embraces national government representatives and associations for producing, distributing, and commercializing products and services. Within the organization and functioning of Icontec, there is the National Standardization Department, which manages standards production nationwide. Its main objective is to support commercial exchange, communication, and technology transfer by producing standards at the national and international levels. The National Standardization Department counts on numerous technical standards committees. These technical committees are committed to elaborating and updating the standards documents assigned to different working groups. In the field of terminology, the work done since the 90s by the GITT coor‐ dinate by Plested-Álvarez, and the efforts made in various universities paved the way to acknowledging how crucial terminology work is for a country’s social and economic development. By this time Colombia was represented at the ISO/ TC37 general meetings. This recognition was key for increasing the relevance of terminology standardization for the country’s interests. These first participations of Colombia delegates started the configuration of a national Terminology work, principles, and applications: Decades of development from Colombia 93 <?page no="94"?> agenda regarding the specific objectives in the international technical commit‐ tee. One of the most outstanding achievements in developing terminology work in Colombia was the creation of the Icontec technical committee 218 with the name terminology and other linguistic and content resources. Along with the experience gained throughout the previous years, international participation, and analysis of the Colombian context, the terminology work started to be related to translation and interpreting activities. At first, it was participating in the project to create an ISO standard for the translation services. As a result, in 2010, Icontec issued the first national standard on translation services and the requirements NTC 5808 to provide these services. Linked to the application of terminology standards and the terminology principles advanced by Budin’s work, it is worth mentioning the doctoral thesis written at the University of Vienna, under the supervision of Budin and Plested-Álvarez. This thesis (Díaz Vásquez 2010) explores the application of International Terminology Standards ISO / TC37 in the export sector in Colom‐ bia, and the WIKO model by Budin. Building on this tradition, Díaz Vásquez continued the tradition of terminology work by supervising the thesis by Casals Rispau (2025) on terminology-based meaning management in multinational corporations. 3 Multilingual Terminology Dictionary Technically, Semantic Business and Vocabulary Rules (SBVR) is a series of spe‐ cifications within the domain of Business Modelling that provide multilingual terminological data under a given structure. SBVR specifications are based on predicate logic, as in terminology theory. As a Content Model, SVBR documents unambiguous, concept-centric, multilingual, and semantically rich capability to define meanings used by agents (humans, machines) in the same industry, discipline, profession, organization, or field of study. Moreover, the SBVR Content Model is also a type of ontology since it highlights the relation between meanings and entities in the world. The SBVR-EAFIT is a business-specific Multilingual Terminology Dictionary (MTD) developed upon the domains of international business taught at the EAFIT University. SBVR’s ultimate purpose is to eliminate ambiguity in busi‐ ness communication, especially in governance documents, thus avoiding or mitigating numerous costly business risks. The foundations of SBVR are rooted in the terminological principles and methods described in ISO/ TC 37, named Language and Terminology, responsible for establishing terminology standards, 94 Juan Carlos Diaz Vásquez, María Cecilia Plested Alvarez <?page no="95"?> as well as in the terminology theory (Budin 2001, 2013, 2019), specifically regarding the central role of the concept in developing concept relations, and their representation. This terminological resource reflects the academic areas of the International Business (IB) at EAFIT University, located in Medellín, Colombia. This is why EAFIT is making part of its name. The making of the SBVR-EAFIT MTD followed terminology principles from the terminology standards, starting with configuring a corpus of basic texts used in IB classes, followed by the time extraction, supported by automated computer systems, and the selection of the terms. Parallel to this work, the terminology modelling establishes generic relations and determines their relations. Finally, the multilingual equivalents (Spanish, French, German, and Japanese) were assigned to the more than 450 terms and definitions in English. Related to this research endeavor, there are implications of terminology ma‐ nagement and knowledge organization for the business context, brought to life in a recent doctoral thesis at EAFIT University (Casals 2025). This thesis explores the characteristics of corporate language beyond its regular understanding as a national language. Among the core elements, this thesis sheds light on specialized corporate terminology grounded in the concept-centred perspective and the centrality of terminology to represent knowledge, information, and communication in multilingual settings. The following paragraphs briefly describe the WIKO model, which is central to the research projects and works previously mentioned. 4 The WIKO model Taking this constant element lingering throughout the above brief revision of the influence of Budin’s work, we present it within the framework of knowledge management as a key factor for the competitive value and the information, communication, and knowledge to find a strong connection with the model of knowledge organization based on terminology principles. The WIKO model (Budin 1996) is in those aspects basically, that is, communication, information and knowledge, a theoretical and methodological approach that can be applied with specific organizations as a way to organize, manage and transfer knowledge based on terminological principles, that can be used in seeking the aim of creating competitive value or that can be integrated to a knowledge-based strategy for different purposes. The WIKO model unites different theoretical elements, as does terminology. Logic is one of the sciences that explain terminology and, therefore, is an Terminology work, principles, and applications: Decades of development from Colombia 95 <?page no="96"?> integral part of this model. This discipline has also been developed from several approaches, including mathematics, formal logic, the theory of knowledge, and epistemology. In the general sense, logic is the theory of good thinking, which considers the act of thought, thought content and objects of thought (Felber 1995). From the perspective of the Philosophy of Science (The Vienna Circle), propositional logic was considered the most important type of logic, however, the conceptual logic is more significant for terminology (Oeser 1992). To cover the aspects that make up the basis of the WIKO model, i.e., knowledge, communication and information, it is necessary to recognize the importance of the ontological structures, i.e., the representation of the real world (or it is conceived like this) and the epistemological dimension (the concepts or abstractions that make part of the cognitive knowledge). Both structures are traversed by the factor of representation, i.e., semiotics, within which the language can be found, and these structures possess features of mediation, that is, transmission. This transmission factor is implicit communication in all levels of knowledge apprehension (observation of reality, formulation of hypotheses, and scientific research methods), and this communication is given in terms of the knowledge management units building up information. Based on the concepts defined as units of thought and based on the terminology as an organizer of the specialized knowledge through systematic structuring of the knowledge units (concepts), Budin proposes the integration of terminological structures (terminology system) as a cross-cutting strategic system, the WIKO model. 4.1 Knowledge Organization In this regard, it is necessary to distinguish clearly between epistemological and cognitive knowledge. The latter occurs in the personal sphere, i.e., of the acts of thinking as perception, which leads to the internalization of concepts. Science has mainly addressed these processes as cognitive psychology (Brunner et al. 1986). This type of knowledge refers to the unconscious processes that operate in individuals, related to their perception of reality. In other words, it is the organization of knowledge at the individual level (Neisser 1987). The other one, the epistemological knowledge, is organized through formulating theories and hypotheses and their confirmation or falsifiability (Popper 2000) in the empirical application. In this case, the sense of organization is assumed to be a structure where static and dynamic aspects are recognized and reflect a classification in the social sphere as made in an organization. Knowledge organization represents a fundamental pillar for this model because it refers to how the apprehension of reality is established and transformed into learning 96 Juan Carlos Diaz Vásquez, María Cecilia Plested Alvarez <?page no="97"?> and knowledge. It is precisely from reflections on reality and object perception that the questions on forming knowledge upon the basis of theories and their development emerge. The units of knowledge, i.e., concepts, which constitute theories and assumptions, are analyzed within a system. Not individually (Budin 1996). This dependence on the context of scientific knowledge is like the situation in quantum physics, where the individual particles can only be considered in a general system (Budin 1996). In other words, separate scientific concepts are meaningless; they only make sense when they are about different ideas, whose relationship is tested in the context of a given theory. From this perspective, the ontological and epistemological elements are combined with the evolutionary and informational aspects of science to understand, how it is proposed by Oeser, science as a system of information in the light of the evolutionary history of mankind with their scientific production processes as hypothesis construction, deduction, induction, inclusion etc. (Oeser 1976). 4.1.1 Information Organization The WIKO model takes the science of communication as the organization of the medial processes of information science and in this sense, terminology is also connected with the field of library science, i.e., with the determination of infor‐ mation sources (Budin 1996). It also touches the field of documentation science, i.e., the scientific verification of the information and its sources. In this way, a connection between knowledge organization and information organization can be established. In contrast, the concept of organization is understood as access to information (Budin 1996). Similarly, information organization is reflected intellectually - the content of the information - and technically - in terms of the management, administration, management, transmission and classification of documents and other types of content, information and data. Another approach to information science connects it with theoretical science, where the former is integrated throughout the general system of scientific work. Knowledge transmission through representations and publications is an integral part of the scientific system. The need to transmit knowledge within appropriate means corresponding to referenced processes and functions generates information processes through this. New ideas are developed through the dynamics of these information processes (representation-reception), and others are revalidated. In other words, new knowledge is created. Accordingly, this dynamic also generates the practical application side of knowledge, that is, techniques and processes that lead to outcomes such as products (Budin 1996). In this way, information management contributes to the expansion and development of science from the intellectual level to the technical level. At the Terminology work, principles, and applications: Decades of development from Colombia 97 <?page no="98"?> same time, it can also be said in the marketplace where information turns into a valuable competitive resource. As noted above, information sciences are linked with documentation science, in the sense of the scientific verification of the information and its sources, which is a primary aspect of the WIKO model. 4.1.2 Communication Organization Communication of specialized information and knowledge makes it an essential part of the WIKO model. Much more when language is understood simultan‐ eously as a system in which different forms of communication are recognized according to the complexity of language in a specific field, in this way, language is also linked to the various levels of organization within the WIKO model, where the concept of organization plays a central role in outlining these levels and their corresponding degrees of complexity. When language is understood as a mechanism in concrete communication conditions, given by the type and nature of concepts generated by itself and conditioned by the specific field of knowledge that represents, Budin proposes an appreciation of language as an open system (Budin 1996). This feature is evident in the interaction of language with reality, in its portrayal of reality and in the influence that it receives from the same (Budin 1996). This reveals one of the principles of the model, which is that meaning is not an emergent feature of the system; it is given, but in their mutual communication with the cognitive knowledge organization (type of knowledge remaining in the individual) the symbiotic interaction with the organization of the epistemological knowledge is what gives meaning to the communication system, similarly, at the level of the organization, which is another feature of this system responding to the principle of functional integration, where complexity increases with each one of the levels of organization. 4.1.3 The terminology as a cross-cutting element of connection in the WIKO model After briefly exploring the knowledge, communication, and information sys‐ tems making up the WIKO model, attention should shift to the terminology and its concept system. Terminology is defined here as an epistemic, informational, and communicative system, in which the terminological unit, as a relation of correspondence between concept and representation, is the primary unit of reference (Budin 1996). From this approach, terminology is an inherent and substantial presence in other spheres of the organization. More than that, it could be said that terminology constitutes the hub of the WIKO model and is 98 Juan Carlos Diaz Vásquez, María Cecilia Plested Alvarez <?page no="99"?> simultaneously the theoretical ground of the present paper, always considering terminology contextualized in systems of concepts and their representations. Terminological analysis is the first point of reflection, as the core instrument in developing the theory of science (Heinish 2023). The main criteria to establish the terminological analysis are the pragmatic determination of the situation to consider (that is, the determination of communicative actions objectives), the target group, and the determination of the context conditions (legislation, economic and social conditions, among others). The second criterion in the terminological analysis is the selection of the degree of information density, which includes decisions on the specialization degree that will have the repre‐ sentations of scientific concepts, determined from the target group (laymen, professionals, or experts) within a communicative context. And the third criterion is the specific organization of communication, in other words, the inclusion of these conceptual representations in a particular language with its rules and conventions. The terminology work is then understood as the semiotic representation of the epistemic and conceptual organization of knowledge through which the organization of communication is concretized. The multi-functional or multidisciplinary nature of terminology is a trait that this discipline shares with the WIKO model, and more than sharing this characteristic, terminology is conceived as a cross-cutting system that connects the other subsystems building up the WIKO model (Budin 1996). Seen from this perspective, the theory of terminology also includes certain features, such as the systematic (as transverse, multi-functional system in the WIKO model), its relationship with reality (ontological character) or its character of evolution (by its mechanisms of development, diversity, stability, etc.) and its universal pragmatic applicability (determination of the system’s purpose). It is essential to highlight that terminology should not be confused with a subsystem of information, communication, or knowledge; it must be understood as a concentration of emerging characteristics from the other interconnected systems. A practical way to visualize this idea is through a Venn diagram, which contains three spheres of communication, information, and knowledge systems, which are intersected, and the resulting intersection is the terminology system (see Fig. 1). Terminology work, principles, and applications: Decades of development from Colombia 99 <?page no="100"?> Fig. 1: Visual representation of the WIKO model (Budin 1996) These aspects are the characteristics of a dynamic system manifested, for example, at the level of the concept and the structure in which it is formed. Another feature of terminology as a dynamic system (which is an effect of using terminology) is its organization, which has materialized through the re‐ duction of complexity in specialized communication by increasing accuracy and stability in the specialized language, which in turn is achieved by terminology standardization. 5 Conclusion The brief overview of the events, actors and developments within the field of terminology that is presented in this contribution evidences the depth of the influence of Budin’s work and ideas in Colombia. Precisely, that was the main objective in collecting and presenting this evidence, and by doing so giving a glimpse into the wide range of fields where terminology and terminology work can be used for communication in specialized contexts. The long history 100 Juan Carlos Diaz Vásquez, María Cecilia Plested Alvarez <?page no="101"?> of collaborative work among different agencies, research groups, universities and government bodies is also a clear indicator of the growing interest in terminology for various purposes. The theoretical and methodological sources on which the evolution of terminology in Colombia is rooted is at the same time deeply connected to the influence of Budin’s work. Therefore, the future of specialized communication and technical writing in the Colombian context refers clearly terminology principles taken from the person that ignited a great deal of these and future developments in these areas. Bibliography Betancurth, Liliana/ Plested, María Cecilia (2021). Comunidades del Delta del Río Atrato, inclusión social a través de la investigación científica [Conference presentation]. XVIII Congreso Colombiano de Geología, Medellín, Colombia. Brunner, Jerome S./ Goodnow, Jacqueline J./ Austin, George A. (1986). A Study of Thin‐ king. New Brunswick: Transaction. Budin, Gerhard (1996). Wissensorganisation und Terminologie: Die Komplexität und Dynamik wissenschaftlicher Informations- und Kommunikationsprozesse. Tübingen: Gunter Narr Verlag. Budin, Gerhard/ Schmitz, Klaus-Dirk (2000). Terminology Management in the Digital Age: Challenges and Opportunities. TermNet News, 1(1), 12-17. Budin, Gerhard (2001). A Critical Evaluation of the State-of-the-Art of Terminology Theory. Terminology Science and Research, 12(1-2), 7-23. Budin, Gerhard (2002). Der Zugang zu mehrsprachigen terminologischen Ressourcen: Probleme und Lösungsmöglichkeiten. In: Franz Mayer/ Klaus-Dirk Schmitz/ Ulrich Zeumer (Eds.). eTerminology. Akten des Symposions, Köln, 12-13 September 2002. Cologne: Deutscher Terminologie-Tag, 185-200. Budin, Gerhard (2013). Multilingual Terminology. In: Peter Kelly (Ed.). Multilingual Terminology. London/ New York: Routledge. 45-67. Budin, Gerhard (2015). Terminology in the Age of Digital Communication. In: John James (Ed.). Terminology in the Age of Digital Communication. London/ New York: Palgrave Macmillan, 89-105. Budin, Gerhard (2019). Zum Entwicklungsstand der Terminologiewissenschaft. In: Drewer, Petra/ Pulitano, Donatella (Eds.). Terminologie: Epochen, Schwerpunkte, Umsetzungen. Berlin: Springer, 7-19. Budin, G., Knowledge Organization and the Role of Terminology in Knowledge Organi‐ zation and the Role of Terminology. Springer, 2010. Terminology work, principles, and applications: Decades of development from Colombia 101 <?page no="102"?> Díaz Vásquez, Juan Carlos (2010). Aplicación de las Normas Terminológicas Internaciona‐ les ISO/ TC37 en el sector exportador colombiano. PhD dissertation. Vienna: University of Vienna. Casals Rispau, Sergi (2025). Managing meaning in Multinational Corporations: A critical realist study of corporate terminology across borders. PhD dissertation. Medellín: Universidad EAFIT. Felber, Helmut (1995). Allgemeine Terminologielehre und Wissenstechnik. Theoretische Grundlage. Vienna: TermNet. Heinisch, Barbara (2023). Terminological usability - Adapting terminological databases to different user groups according to usability principles: The case of UniVieTerm. In: Bhreathnach, Úna/ Nissilä, Nina/ Velicu, Andreea (Eds.). Terminology Science & Research / Terminologie: Science et Recherche, 26, 24-44. ISO 704 (2022). Terminology work - Principles and methods. Geneva: International Organization for Standardization ISO/ FDIS 860 (2007). Terminology work - Harmonization of concepts and terms. Geneva: International Standardization Organization. ISO/ WD 29383 (2002). Terminology Policies. Geneva: International Standardization Organization. Neisser, Ulrich (1987). From direct perception to conceptual structure. In: Neisser, Ulrich (Ed.). Concepts and conceptual development. Ecological and intellectual factors in categorization. Cambridge: Cambridge University Press, 11-24. Oeser, Erhard (1992). Terminology and Philosophy of Science. In: Draskau, J./ Picht, Heribert (Eds.). International Conference on Terminology Science and Terminology Planning/ International IITF-Workshop on Theoretical Issues of Terminology Science. IITF Series 4. Wien: TermNet, 24-34. Oeser, Erhard (1976). Wissenschaft und Information. Band III. Struktur und Dynamik erfahrungswissenschaftlicher Systeme. München/ Wien: Oldenbourg. Plested Álvarez, María Cecilia (2010). Una mirada de conjunto. Íkala, revista de lenguaje y cultura, 15(25), 9-17. Plested Álvarez, María/ Casals, Sergi/ Vallejo, Gloria (2010). Un enfoque onomasiológico de la Ciencia. V Coloquio Internacional sobre la Historia de los lenguajes Ibero-román‐ icos de especialidad: Comunicación y transmisión del saber entre lenguas y culturas. Leipzig: Universidad de Leipzig. Plested, María Cecilia/ Giraldo, Blanca Stella (2000). Normalización terminológica para el Sistema Nacional de Ciencia y Tecnología. In: VII Simposio Iberoamericano de Terminología. Lisboa: Riterm, 821-830. Popper, Karl Raimund (2000). Lesebuch: Ausgewählte Texte zur Erkenntnistheorie, Philo‐ sophie der Naturwissenschaften, Metaphysik, Sozialphilosophie. Tübingen/ Stuttgart: Mohr Siebeck. 102 Juan Carlos Diaz Vásquez, María Cecilia Plested Alvarez <?page no="103"?> Wright, Sue Ellen/ Budin, Gerhard (1997). Handbook of Terminology Management: Volume 1: Basic Aspects of Terminology Management. Amsterdam/ Philadelphia: John Benjamins. Wüster, Eugen (1985). Einführung in die allgemeine Terminologielehre und terminolo‐ gische Lexikographie. 2. Aufl. Kopenhagen: Fachsprachliches Zentrum Handelshoch‐ schule. Wüster, Eugen (1979). Einführung in die allgemeine Terminologielehre und terminolo‐ gische Lexikographie. Wien/ New York: Springer-Verlag. Terminology work, principles, and applications: Decades of development from Colombia 103 <?page no="105"?> Einflussfaktoren bei der Terminologiearbeit im Recht Elena Chiocchetti, Natascia Ralli Abstract: Der Beitrag untersucht, wie sich drei wichtige Faktoren (Fach‐ gebiet, Zielsetzung und Nutzergruppen) auf die Terminologiearbeit aus‐ wirken. Am Beispiel des Rechts wird gezeigt, wie diese Faktoren die Datenmodellierung, die Datenstruktur und die Auswahl der Datenkatego‐ rien einer Terminologiedatenbank entsprechend den ISO-Normen und dem DatCatInfo-Repository beeinflussen können. Das Erfassen verschie‐ dener Varietäten derselben Sprache, die verschiedenen Anwendungssze‐ narien und Anforderungen unterschiedlicher Zielgruppen müssen bei der Konzeption und (Um)Strukturierung einer Terminologiedatenbank berücksichtigt werden. Dabei gilt es, unter Wahrung der terminologischen Grundsätze mitunter einen tragfähigen Kompromiss zwischen den ver‐ schiedenen Einflussfaktoren zu finden, da Terminologiearbeit stets zweck- und nutzerorientiert erfolgen sollte. 1 Einleitung Die ISO-Norm 1087 (2019: 3.5.1) definiert Terminologiearbeit als die systema‐ tische Sammlung, Beschreibung, Bearbeitung und Darstellung von Begriffen und deren Bezeichnungen. Ziel ist es, Wissen in einer oder mehreren Sprachen darzustellen und zu vermitteln. In dieser Hinsicht bildet gut recherchierte und qualitativ hochwertige Terminologie die Grundvoraussetzung für korrekte, konsistente und transparente Fachkommunikation sowie effizientes Wissens‐ management in Institutionen und Unternehmen (RaDT 2015: 4). Allerdings erfordert das Erstellen terminologischer Einträge ein fundiertes Wissen über die Grundlagen der Terminologielehre sowie über die Grundsätze und Methoden der Terminologiearbeit. Gemäß diesen Grundsätzen und Metho‐ den wird Terminologie begriffsorientiert erfasst. Das heißt, terminologische Einträge enthalten alle Informationen zu einem Begriff wie Benennungen, <?page no="106"?> Definition, Beispielkontexte, Anmerkungen usw. Diese Informationen werden Datenkategorien zugeordnet und auf drei Ebenen verwaltet (ISO 26162-1 2019: 4.2; ISO 30042 2019: 8.2-8.3): - Auf der Begriffsebene sind Informationen vorhanden, die den Begriff als Ganzes betreffen, wie Angaben zum Fachgebiet oder Projekt. Auf dieser Ebene kann auch die Definition erfasst werden, vorausgesetzt, sie ist nicht auf der Sprachebene verankert (siehe Abschnitt 2). - Auf der Sprachebene befinden sich die Sprachen, in denen die Terminologie erfasst ist, und die Bezeichnungen zum Begriff (einschließlich Benennun‐ gen, Eigennamen und Symbole) sowie sämtliche sprachspezifische Informa‐ tionen wie sprachbezogene Anmerkungen (Drewer und Schmitz 2017: 132). Dieser Beitrag legt den Schwerpunkt auf die sprachlichen Bezeichnungen, insbesondere auf die Benennungen. Auf der Sprachebene kann auch die Definition erfasst werden, sofern sie nicht auf der Begriffsebene angelegt ist (siehe Abschnitt 2.1). - Auf der Benennungsebene sind alle Informationen zu den einzelnen Benen‐ nungen wie Grammatik, geografischer Gebrauch, Kontextbeispiele usw. zu finden. Diese Ebene dient dazu, den Grundsatz der Benennungsautonomie zu gewährleisten: Alle Benennungen sind als eigenständige Teileinheiten anzusehen, die „jeweils mit einer Reihe von (abhängigen) Datenkategorien dokumentiert werden“ (Drewer und Schmitz 2017: 131). Die praktische Umsetzung des Grundsatzes der Begriffsorientierung unter Berücksichtigung der unterschiedlichen Ebenen ist nur die Spitze des Eisbergs. Es gibt nämlich Faktoren, die sich auf das Erfassen, das Beschreiben und die Darstellung von Terminologie auswirken: das Fachgebiet, die Zielsetzung der Terminologiesammlung und die Nutzergruppe. Diese beeinflussen die Datenmodellierung, die Datenstruktur und die Auswahl der Datenkategorien entsprechend der Norm ISO 12620-1 (2022) und dem international anerkannten Data Category Repository DatCatInfo (Warburton und Wright 2019). Anhand des Rechts soll der vorliegende Beitrag diese Einflussfaktoren be‐ leuchten, die einzeln in Bezug auf die Datenstruktur und die Auswahl der Da‐ tenkategorien behandelt werden. Dabei wird die Terminologiearbeit zu einem anwendungsorientierten und praxisbezogenen Forschungsgegenstand (Budin 2019: 8). Im Folgenden widmet sich Abschnitt 2 gezielt dem Einflussfaktor des Fachgebiets, mit besonderem Augenmerk auf die Herausforderungen, die beim Systematisieren kulturgebundener Fachgebiete wie dem Recht auftreten. Abschnitt 3 behandelt die unterschiedlichen Zielsetzungen, denen eine Termi‐ nologiesammlung dienen kann. Darauf aufbauend befasst sich Abschnitt 4 106 Elena Chiocchetti, Natascia Ralli <?page no="107"?> mit den Erwartungen der wichtigsten Nutzergruppen, die ebenfalls Einfluss auf die Terminologiearbeit und die Datenstruktur nehmen. Abschnitt 5 fasst abschließend die zentralen Aspekte zusammen. 2 Das Fachgebiet / Fachgebiet/ ist eine Datenkategorie, die einen Begriff einem bestimmten Be‐ reich des menschlichen Wissens zuweist (Drewer und Schmitz 2017: 115). Sie ist üblicherweise auf Begriffsebene angesiedelt (ISO/ TS 24634 2024: 4.1), da sie sich auf den Begriff als Ganzes bezieht. Die Abgrenzung des Fachgebiets stellt eine der wichtigsten Phasen der Terminologiearbeit dar. Diese Tätigkeit umfasst mehrere Schritte und erfolgt für jede Sprache, in der die terminologischen Daten zu erfassen sind. Auf Grund‐ lage des Fachgebiets wird einschlägiges Dokumentationsmaterial gesammelt. Anschließend werden Begriffe beschrieben, voneinander abgegrenzt und in ein Begriffssystem eingeordnet. Es werden auch die Benennungen zum Begriff identifiziert und mit zusätzlichen Informationen versehen. Alle Informationen werden in Datenkategorien erfasst, deren Auswahl stark mit dem betreffenden Fachgebiet zusammenhängt. In einer Terminologieda‐ tenbank im Bereich Biologie wird es zum Beispiel unter der Datenkatego‐ rie / Benennungstyp/ Auswahllistenwerte wie „wissenschaftlicher Name“ und „Trivialname“ geben. Für eine Datenbank im Bereich Chemie wäre hingegen ein Auswahllistenwert wie „Symbol“ sinnvoll. Die Datenkategorien / Bild/ oder / Grafik/ können in Fachgebieten wie Zoologie und Maschinenbau sehr nützlich sein, aber wahrscheinlich weniger in Fachgebieten wie Philosophie und Recht. Das Fachgebiet dient auch dazu, gleichlautende Benennungen zu unterschei‐ den (Drewer und Schmitz 2017: 115). Vereinfacht ausgedrückt, bezeichnet zum Beispiel Blatt einen Teil einer Pflanze in der Botanik und ein Stück Papier im Druckwesen. Durch die unterschiedlichen Angaben in der Datenka‐ tegorie / Fachgebiet/ können in Terminologiesammlungen solche Begriffe mit gleichlautenden Benennungen sehr einfach getrennt werden. Im Rahmen der mehrsprachigen Terminologiearbeit ist ein zusätzlicher As‐ pekt zu beachten: Handelt es sich um ein Fachgebiet, das denselben kognitiven Hintergrund hat oder internationalisiert ist (Sandrini 1996a: 150-151), kann die Definition auf der Begriffsebene angesiedelt werden - vorausgesetzt, es wird eine Pivot-Sprache festgelegt. Dies ist eine im Voraus definierte Referenz‐ sprache, die als Bezugspunkt für die Anordnung der Benennungen in den verschiedenen Sprachen innerhalb eines terminologischen Eintrags dient. In solchen Fachgebieten (z. B. Mathematik, Physik) sind viele Begriffe interna‐ Einflussfaktoren bei der Terminologiearbeit im Recht 107 <?page no="108"?> tional anerkannt. Beispielsweise ist die Kreiszahl Pi (π) eine mathematische Konstante, die weltweit anerkannt ist (Britannica), sodass eine Definition auf der Begriffsebene formuliert werden kann. Demzufolge gilt die Definition in allen Ländern, unabhängig von Sprache und Kultur. Eine gemeinsame und internatio‐ nal anerkannte Definition bedeutet vollständige begriffliche Übereinstimmung, was zu einer formalen Äquivalenz der Benennungen in den verschiedenen Sprachen führt (Arntz et al. 2021: 146). Was passiert aber, wenn ein Fachgebiet nicht internationalisiert ist oder einen unterschiedlichen kognitiven Hintergrund hat, wie das Recht? Der nächste Abschnitt geht auf diese Frage ein. 2.1 Der Rechtsbereich Jedes Rechtssystem hat eigene Begriffs und Wissensstrukturen, die durch historische, kulturelle und gesellschaftliche Entwicklungen beeinflusst werden (Sandrini 1996a: 149; Šarčević 1997: 232). Die daraus resultierende Rechtstermi‐ nologie gehört zugleich einem Sprach- und einem Rechtssystem an (Gambaro und Sacco 2024: 8), denn das Recht wird durch die Rechtssprache sowohl beschrieben als auch geschaffen. Daher ist die Rechtsterminologie fest im jeweiligen Rechtssystem verwurzelt und somit systemgebunden (Cao 2007: 23-24; de Groot 1999: 12). Nehmen wir Stalking als Beispiel. Das beabsichtigte und wiederholte Verfol‐ gen oder Belästigen einer Person, sodass ihre Sicherheit bedroht und sie in ihrer Lebensgestaltung erheblich beeinträchtigt wird, wird je nach Rechtssystem an‐ ders behandelt. In Rumänien wird das Stalking bislang nicht als eigenständiger Straftatbestand betrachtet (EIGE). In der Slowakei hingegen fällt das Stalking unter den Tatbestand der häuslichen Gewalt (EIGE). In Italien und in Österreich wird diese Verhaltensweise als eigenständiger Straftatbestand geahndet, der über den Tatbestand der häuslichen Gewalt hinausgeht (bistro). Dies ist nur ein Beispiel, um zu verdeutlichen, dass vollständige begriffliche Übereinstimmung zwischen Rechtsbegriffen aus unterschiedlichen Rechtssys‐ temen eher selten anzutreffen ist (Cao 2007: 29; de Groot 1999: 21; Sandrini 2014: 149). Aus dieser Sicht muss der Grundsatz der begrifflichen Übereinstimmung bei der Terminologiearbeit mit mehreren Rechtssystemen zwangsläufig relati‐ viert werden (Arntz et al. 2021: 162). Die Systemgebundenheit wirkt sich auch dort aus, wo unterschiedliche Rechtssysteme dieselbe Sprache benutzen. Eine gemeinsame Sprache bedeutet nicht unbedingt, dass dieselbe Rechtskultur und -sprache gänzlich geteilt wer‐ den. Nehmen wir unser bereits genanntes Beispiel des Stalkings. Im deutschen 108 Elena Chiocchetti, Natascia Ralli <?page no="109"?> Sprachraum wird diese Verhaltensweise je nach Rechtssystem unterschiedlich bezeichnet (bistro): Nachstellung in Deutschland und in der Schweiz, beharrli‐ che Verfolgung in Österreich und Verfolgungshandlungen in Italien (Südtirol). Innerhalb der deutschen Sprache gibt es so viele Rechtssprachen wie es Rechts‐ systeme gibt, die diese Sprache verwenden. Schließlich gibt es einen weiteren Aspekt, der zu beachten ist. Wie jede natürliche Sprache unterliegt auch die Rechtssprache „dem Prozess des Sprach‐ wandels […], der mit dem Wandel der Welt und der Menschen im Zusammen‐ hang steht“ (Siewer-Kowalkowska 2017: 51). Wenn beispielsweise eine Regelung nicht mehr den Anforderungen der Gesellschaft entspricht, wird sie aufgehoben oder nicht mehr angewandt. Der gesellschaftliche Wandel führt zu neuen Begriffen sowie zu Änderungen bestehender Begriffe bzw. Benennungen und ist in jedem Rechtssystem unterschiedlich. Dies zeigt sich anhand von drei Beispielen. Das erste Beispiel veranschaulicht die Entstehung eines neuen Begriffs: Im Jahr 2016 hat Italien die eingetragene Lebenspartnerschaft für gleichgeschlechtliche Paare (Gesetz 76/ 2016) eingeführt. Somit sind auch neue Begriffe entstanden, die für die Regelung dieser sozialen Bindung erforderlich sind (z. B. in eingetragener Lebenspartnerschaft als Fami‐ lienstand). Das zweite Beispiel verdeutlicht eine Änderung des Begriffs: Seit dem 1. Januar 2019 können in Österreich gleichgeschlechtliche Paare eine Ehe eingehen (VfGH 04.12.2017, G 258-259/ 2017-9). Damit wurde der Begriff Ehe erweitert, da vor der Reform nur verschiedengeschlechtliche Paare heiraten konnten. Das dritte Beispiel zeigt die Änderung einer Benennung. Mit der Reform des italienischen Familienrechts (Gesetzesvertretendes Dekret 154/ 2013, Art. 105) wurde infolge der Gleichstellung ehelicher und nichtehelicher Kinder nichteheliches Kind durch Kind oder außerhalb der Ehe geborenes Kind ersetzt. Infolge dieses komplexen Bildes stellt sich die Frage, wie die Systemge‐ bundenheit und der Rechtswandel bei mehrsprachiger und rechtssystemüber‐ greifender Terminologiearbeit beschrieben und dargestellt werden können. Aufgrund der oben erwähnten relativen Übereinstimmung der Rechtsbegriffe (Arntz et al. 2021: 162; Sandrini 1996a: 198) kann die Definition nicht auf der Begriffsebene angelegt werden. In diesem Fall muss sie auf der Sprachebene an‐ gesiedelt werden, damit je Rechtssystem eine eigene Definition erfasst werden kann. Durch rechtssystembezogene Definitionen werden kleinere Unterschiede auf der Ebene des Rechtsbegriffs sowie dessen Regelung und Anwendung auf‐ gezeigt. Diese benutzer- und anwendungsorientierte Vorgehensweise erfordert, dass terminologische Einträge die Ergebnisse der rechtsvergleichenden Analyse enthalten, wobei diese nicht in Form von Äquivalenten oder Nicht-Äquiva‐ lenten, sondern als unterschiedliche Grade der Vergleichbarkeit darzustellen Einflussfaktoren bei der Terminologiearbeit im Recht 109 <?page no="110"?> sind (Sandrini 1996b: 9; 2018: 519). Auf diese Weise lassen sich möglichst detaillierte Informationen zu den Begriffen und Benennungen der untersuchten Rechtssysteme bereitstellen, um Nutzer: innen beispielsweise bei der Auswahl der passenden Benennungen zu unterstützten (siehe Abschnitt 4). Befasst sich Terminologiearbeit mit plurizentrischen Sprachen, muss die Terminologiedatenbank so gestaltet werden, dass Sprachvarietäten mitberück‐ sichtigt werden und dabei der Grundsatz der Benennungsautonomie (siehe Abschnitt 1) gewahrt bleibt. Im Recht ist die Sprachvarietät rechtssystemge‐ bunden. Das bedeutet beispielsweise, dass das österreichische Deutsch dem österreichischen Rechtssystem und das Südtiroler Deutsch dem italienischen Rechtssystem zugeordnet ist. Aufgrund unterschiedlicher Faktoren (z. B. Größe der Terminologiedatenbank, (Nicht)Vorhandensein einer Sprachkennung) kann die Sprachvarietät (Ralli 2025): - auf der Sprachebene erfasst werden. Idealerweise sollte der Sprachvarietät eine Sprachkennung (z. B. de-AT, de-CH) gemäß ISO 639 (2023) zugeordnet sein, um einen reibungslosen Datenaustausch zu ermöglichen. - auf der Benennungsebene erfasst werden. In diesem Fall wird sie als untergeordnete Datenkategorie der Benennung dargestellt, wobei im Hin‐ blick auf den Datenaustausch und die Interoperabilität die Sprach- oder Länderkürzel nach ISO 639 (2023) oder ISO 3166-1 (2020) (z. B. AT, DE, CH) verwendet werden sollten. Die Datenkategorie kann / geografischer Gebrauch/ oder / Rechtssystem/ genannt werden. Die Datenkategorie / geografischer Gebrauch/ könnte auch im folgenden Fall Anwendung finden: In Unterfachgebieten, die meist auf subnationaler Ebene geregelt werden (z. B. Baurecht), können unterschiedliche Benennungen aus verschiedenen Bundesländern, Kantonen, Regionen oder Provinzen durch diese Datenkategorie dem richtigen regionalen Sprachgebrauch zugeordnet werden. Für einen reibungslosen Datenaustauch wäre es angebracht, die von der ISO-Norm 3166-2 (2020) vorgesehenen Kürzel zu verwenden (z. B. DE-BY für Bayern). Hinweise auf Inkongruenzen zwischen zwei oder mehreren Begriffen können in einer / Anmerkung/ auf der Begriffsebene erfasst werden (siehe Abschnitt 3). Solche Anmerkungen haben jedoch noch weitere Funktionen und können Informationen verschiedener Art enthalten. Sie können terminologische Lücken aufzeigen sowie Eigenheiten eines Rechtsbegriffs hervorheben, die nicht Teil der Definition sind. In Anmerkungen können auch Informationen zu Begriffs‐ änderungen einfließen. Informationen zu Benennungsänderungen können in einer / Anmerkung/ auf Benennungsebene erfasst werden. 110 Elena Chiocchetti, Natascia Ralli <?page no="111"?> Da das Recht ein sehr umfassendes Fachgebiet ist, ist es angeraten, Unterfach‐ gebiete (z. B. Strafrecht, Familienrecht, Sachenrecht) zu definieren, um etwaige gleichlautende Benennungen zu unterscheiden. Ein Beispiel hierfür ist die Benennung Handlung, die für unterschiedliche Begriffe steht: ein menschliches Verhalten im Zivilrecht, ein willensgetragenes Verhalten im Strafrecht und einen Gegenstand der Leistung im Schuldrecht (bistro). 3 Zielsetzung Das Anlegen einer Terminologiesammlung kann verschiedenen Zwecken die‐ nen, oft auch mehreren zugleich. Häufig sollen die ein- und mehrsprachige technische Kommunikation, die Translation (das Übersetzen und Dolmetschen) oder die Sprachplanung in Unternehmen bzw. für Sprachgemeinschaften unter‐ stützt werden. Der Einsatz für Lehre oder Wissensorganisation kann ebenfalls ein primäres oder sekundäres Ziel einer Terminologiesammlung sein. Dabei wird je nach Fall ein deskriptiver oder präskriptiver Ansatz verfolgt. Die Zielsetzungen einer Terminologiesammlung spiegeln sich in deren Da‐ tenstruktur und in der Auswahl der Datenkategorien wider. Eine zentrale Aufgabe bei der Konzeption von Terminologiedatenbanken besteht darin, einen ausgewogenen Kompromiss zwischen den Anforderungen zu finden, die sich aus etwaigen unterschiedlichen Zielsetzungen ergeben. Datenstruktur und Inhalte sind bei Bedarf so zu verwalten, dass sie den geplanten kommunikativen (z. B. Verfassen oder Übersetzen von Texten) bzw. kognitiven (z. B. Informati‐ onssuche, Klärung begrifflicher Unklarheiten) Zielen gerecht werden (Tarp 2017: 123-124). Für Redaktions- und Translationszwecke sind beispielsweise Datenkategorien wie / Grad der Synonymie/ oder / Grad der Äquivalenz/ nütz‐ lich. Bei Terminologiesammlungen für das Dolmetschen können Informationen zur / Aussprache/ wichtig sein. Steht hingegen der einheitliche Gebrauch der Unternehmensterminologie im Vordergrund, sind Datenkategorien wie / Status/ mit Auswahllistenwerten wie „bevorzugt“, „zulässig“ und „verboten“ relevant. Bei nationalen Terminologiedatenbanken mit einem sprachplanerischen An‐ satz können Angaben zur / Häufigkeit/ von Benennungen die Sprachnormung steuern. Dient eine Sammlung primär der Wissensorganisation, können ent‐ sprechende Datenkategorien für Begriffssysteme oder Ontologien eingesetzt werden. Die Strukturierung des Wissens, das in eine Terminologiesammlung einfließt, ist ein durchwegs positives Ergebnis der Terminologiearbeit. Dabei sind eine saubere Datenmodellierung und Datenstruktur gemäß den terminologischen Grundsätzen sowie die Auswahl der Datenkategorien nach der ISO-Norm Einflussfaktoren bei der Terminologiearbeit im Recht 111 <?page no="112"?> 12620-1 (2022) und DatCatInfo in Terminologiedatenbanken unverzichtbar. Nur so können der Austausch und die Wiederverwendung von terminologischen Daten sowie die Interoperabilität von verschiedenen Anwendungen gewähr‐ leistet werden. Ein Ziel kann nämlich auch darin bestehen, terminologische Daten für KI-Anwendungen verwendbar zu machen, beispielsweise, um sie als Linked Data oder als Wissensgraphen in anderen (sprachtechnologischen) Tools zu nutzen (vgl. u.-a. Chiarcos et al. 2013; Di Buono et al. 2020; DIN SPEC 91526 2025; Montiel-Ponsoda et al. 2015). Terminologiesammlungen mit rechtlichem Inhalt können verschiedene Zwe‐ cke verfolgen, insbesondere die Unterstützung der internationalen juristischen Kommunikation, der ein- oder mehrsprachigen Rechtsetzung, der Rechtsüber‐ setzung, des Gerichtsdolmetschens oder der Sprachplanung. Dabei ist es un‐ erlässlich, dass Terminologieressourcen infolge von Gesetzesreformen und -novellierungen regelmäßig aktualisiert werden. Angesichts der zunehmenden Globalisierung des Rechts und den vielen Versuchen, Regeln und Gesetze auf internationaler Ebene zu harmonisieren (Gambaro und Sacco 2024: 27-28), können rechtsterminologische Datenbanken einen entscheidenden Beitrag zur internationalen Zusammenarbeit und Kommunikation leisten. Auch innerhalb derselben natürlichen Sprache kann es unterschiedliche Rechtssprachen geben (siehe Abschnitt 2.1), sodass eine rechtsterminologische Sammlung mit mehre‐ ren Rechtssystemen nicht zwingend mehrsprachig sein muss (vgl. Muhr und Peinhopf 2015). Bei rechtsterminologischen Sammlungen hängt die Auswahl der erfor‐ derlichen Datenkategorien ebenfalls vom Zweck der jeweiligen Terminolo‐ gieressource ab. Zum Beispiel kann eine Angabe zum / Eintragstyp/ wie „phraseologischer Eintrag“ oder zum / Benennungstyp/ wie „Phraseologie“ als Auswahllistenwert in translationsorientierten Terminologiedatenbanken erwünscht sein, da feste Wortverbindungen bzw. Kollokationen im Recht eine besondere Herausforderung für die Translation darstellen können (Biel 2014). Die Angabe des / Status/ einer Benennung ist hingegen für die Sprachplanung im Kontext von National- und Minderheitensprachen mit einem präskriptiven Ansatz von Bedeutung (vgl. Ralli und Andreatta 2018: 19). Der wichtigste Einflussfaktor ist allerdings die Anzahl der berücksichtigten Rechtssysteme (Chiocchetti et al. 2023: 461-463) und damit einhergehend die Frage, ob ein rechtsvergleichendes Ziel verfolgt wird. Wird nur ein Rechtssys‐ tem berücksichtigt, das möglicherweise mehrere Amtssprachen hat (z. B. die Europäische Union), oder werden zwei oder mehrere Rechtssysteme berücksich‐ tigt, die die gleiche Sprache oder unterschiedliche Sprachen verwenden (z. B. Deutschland und Österreich bzw. Deutschland und Frankreich)? Im ersten Fall 112 Elena Chiocchetti, Natascia Ralli <?page no="113"?> ist eine Rechtsvergleichung nicht relevant, im zweiten hingegen unverzichtbar. Wie in Abschnitt 2.1 bereits erläutert, besteht selten völlige Übereinstimmung zwischen Rechtsbegriffen aus verschiedenen Rechtssystemen. Informationen zu Unterschieden in Datenkategorien wie / Grad der Äquivalenz/ oder in einer rechtsvergleichenden / Anmerkung/ können die Arbeit von Rechtsfachleuten ebenso wie jene von Translator: innen erheblich erleichtern. Eine / Anmerkung/ kann dabei auf Begriffs-, Sprach- oder Benennungsebene verortet werden, vor allem abhängig davon, ob sie rechtssystemübergreifende (rechtsvergleichende) oder rechtssysteminterne Informationen enthält. In rechtssystemübergreifen‐ den Terminologieressourcen wird zudem eine Datenkategorie wie / geografi‐ scher Gebrauch/ wichtig, um in einem terminologischen Eintrag den Sprach‐ gebrauch in unterschiedlichen Rechtssystemen mit gleicher Amtssprache zu unterscheiden (siehe Abschnitt 2.1). Im Hinblick auf die Interoperabilität mit anderen Anwendungen wurden bereits mehrere Terminologiesammlungen mit rechtlichem Inhalt in maschi‐ nenlesbaren Formaten wie Linked Data (Cimiano et al. 2015; Rodriguez-Doncel et al. 2018) und Wissensgraphen (Martín-Chozas 2019; Wetzel et al. 2021) aufbereitet, was deren Nutzen für die Zukunft erheblich steigert. In solchen Fällen ist eine Datenmodellierung entsprechend terminologischen Grundsätzen sowie die Auswahl klar abgegrenzter, gut beschriebener und idealerweise in einem international anerkannten Data Category Repository enthaltenen Datenkategorien besonders wichtig. 4 Nutzergruppe Je nach Fachgebiet und Zielsetzung einer Terminologiesammlung können un‐ terschiedliche Nutzergruppen angesprochen werden. Diese haben oft ungleiche Hintergrundkenntnisse und Bedürfnisse, was sich zum Beispiel im Vergleich zwischen Expert: innen eines bestimmten Faches und professionellen Sprach‐ mittler: innen zeigt. Im Fachgebiet Chemie ist für Erstere etwa die Angabe der chemischen Formel einer Verbindung vermutlich wichtiger als für Transla‐ tor: innen, die in erster Linie an der fremdsprachigen Entsprechung des Namens der Verbindung interessiert sind. Typische Nutzergruppen ein- und mehrsprachiger Terminologiesammlun‐ gen sind Personen mit professionellen Sprachkenntnissen und Aufgaben im Bereich der technischen Kommunikation und der Translation (z. B. techni‐ sche Redakteur: innen, Fachübersetzer: innen). Hinzu kommen meist Personen mit Fachkenntnissen im jeweiligen Tätigkeitsbereich (z. B. Ingenieur: innen, Maschinenbautechniker: innen). Weitere Zielgruppen mit unterschiedlichen In‐ Einflussfaktoren bei der Terminologiearbeit im Recht 113 <?page no="114"?> teressen und Vorkenntnissen können ebenfalls direkt oder indirekt zu den Nutzergruppen einer Terminologiesammlung zählen, etwa alle Mitarbeitenden eines Unternehmens. Mit anderen Worten sind die wichtigsten Nutzergruppen in der Regel „subject-field experts, semi-experts, and laypersons“ (Nielsen 2023: 435). Richtet sich eine Terminologiesammlung an unterschiedliche Nutzergrup‐ pen, kann es eine beachtliche Herausforderung werden, jeder Gruppe genau das zu bieten, was sie braucht und sucht. Es ist daher empfehlenswert, die Erwartungen und Anforderungen jeder anvisierten Gruppe im Vorfeld zu erhe‐ ben (Nielsen 2023: 435-437), um die Terminologiedatenbank zielgruppengerecht anzulegen und die Auswahl der Datenkategorien zu steuern. Unterschiedlichen Zielgruppen kann bei Bedarf ein selektiver Zugriff auf bestimmte Daten und Da‐ tenkategorien gewährt werden. So wurden etwa für die Terminologieressource im medizinischen Bereich TriMed (Vezzani et al. 2018) drei Nutzergruppen im Voraus definiert: ärztliches Fachpersonal, Patient: innen sowie Fachleute aus den Bereichen technische Kommunikation und Übersetzung. Jede Gruppe erhält über ein maßgeschneidertes Interface Zugriff auf den für sie relevanten Teil der verfügbaren Informationen. Inzwischen können auch Maschinen und verschiedene (sprachtechnologi‐ sche) Anwendungen als Nutzerinnen terminologischer Daten betrachtet wer‐ den (Chiocchetti et al. 2023: 474). Für diese gelten ebenfalls besondere Anforde‐ rungen wie die Verfügbarkeit der Daten in einem maschinenlesbaren Format. Für Terminologiedatenbanken im Bereich Recht werden traditionell drei Nutzergruppen mit unterschiedlichen Anforderungen identifiziert: Rechtsfach‐ leute, Sprachmittler: innen sowie die breite Öffentlichkeit (Sandrini 2014: 144). Laut Peruzzo (2018: 97-99) suchen Rechtsfachleute vor allem Informationen hin‐ sichtlich der Zuverlässigkeit und letzten Aktualisierung der Daten, Definitionen und (definitorische) Kontexte, eine genaue Unterscheidung nach Unterfachge‐ biet und Rechtssystem, rechtsvergleichende Anmerkungen sowie Angaben und Links zu amtlichen Quellen. Wer hingegen im Rechtsbereich dolmetscht oder übersetzt, möchte in der Regel die Bedeutung eines Begriffs verstehen, wünscht sich Definitionen und Kontexte in mehreren Sprachen, sucht fremdsprachliche Äquivalente und Übersetzungsvorschläge, möchte Synonyme, Abkürzungen und Akronyme finden und braucht Phraseologie (Durán Muñoz 2012: 144). Für Rechtsfachleute könnten daher ein / Zuverlässigkeitscode/ , Angaben über die / letzte Änderung/ und der Hinweis, dass ein Begriff in einem Rechtstext definiert ist (z. B. durch einen entsprechenden Auswahllistenwert), wichtiger sein als Angaben zum / Benennungstyp/ wie „Synonym“ oder „Kurzform“. Solche sprachbezogenen Informationen sind Rechtsfachleuten oft bereits be‐ 114 Elena Chiocchetti, Natascia Ralli <?page no="115"?> kannt, während eine Angabe zur Synonymie zwischen Benennungen oder die Auflösung einer Kurzform (z. B. eines Akronyms) für die korrekte Verdol‐ metschung eines Anwaltsgesprächs ausschlaggebend sein können. Ebenfalls können Angaben zum / Sprachregister/ einer Benennung zu einer passenderen Übersetzung führen. Umgekehrt können Rechtsfachleute ihrerseits Informa‐ tionen wünschen, die für Übersetzer: innen überflüssig sind, wie Angaben zur / Aussprache/ fremdsprachiger Wörter (Peruzzo 2018: 100). Aus diesem Grund gibt es unterschiedliche Vorschläge für die Datenmodellierung und die Datenstruktur, die gezielt die Bedürfnisse von Rechtsübersetzer: innen erfüllen (Bestué 2019: 140-141), während andere sich stärker an den Anforderungen von Rechtsfachleuten (Peruzzo 2018: 101-102) richten. Nicht nur die Auswahl der Datenkategorien, sondern auch deren Inhalte können sich mitunter nach den Bedürfnissen der wichtigsten Nutzergruppe richten. Rechtsfachleute schätzen beispielsweise oft Legaldefinitionen und De‐ finitionen aus Fachquellen, während Sprachmittler: innnen meist einfachere und zugänglichere Definitionen vorziehen (Peruzzo 2018: 102-103; Vanden Bulcke und De Groote 2016: 27). Die dritte Nutzergruppe sind Nichtfachleute, oft alle Bürger: innen eines (mehrsprachigen) Staates oder einer Region, in der eine Minderheitensprache offiziell anerkannt ist. Das Recht betrifft uns alle in zahlreichen alltäglichen Situationen: von der Geburt bis zum Tod, von der Ausbildung bis zur Arbeit, vom Wohnungskauf bis zu Schadensersatzklagen. Rechtliche Begriffe und Benennungen sind somit nicht nur für Gesetzgebung, Rechtsprechung und Verwaltung relevant. Viele kommunikative Situationen betreffen direkt oder indirekt alle Bürger: innen. Letztere können daher ebenfalls Informationen in Terminologiedatenbanken suchen. In der Regel wünscht sich diese Nutzer‐ gruppe laienverständliche Erklärungen von Rechtsbegriffen. Oft sucht sie den Zugang zu terminologischen Einträgen über Benennungen, die nicht in der Fachsprache, sondern in der Allgemeinsprache und in den Medien gebraucht werden (z. B. Schwarzfahren vs. Fahren ohne gültige Fahrkarte, vgl. Danzl 2024). Für solche User können etwa Angaben zum / Sprachregister/ besonders hilfreich sein. 5 Schluss In diesem Beitrag wurden - am Beispiel des Rechts - die wichtigsten Einfluss‐ faktoren bei der Terminologiearbeit untersucht. Dazu zählen im betreffenden Beispiel die Systemgebundenheit von Rechtsterminologie, die Zielsetzung einer rechtsterminologischen Sammlung sowie deren anvisierte Nutzergruppen. All Einflussfaktoren bei der Terminologiearbeit im Recht 115 <?page no="116"?> diese Faktoren wirken sich auf die Datenmodellierung und Datenstruktur einer Terminologiesammlung sowie auf die Auswahl der Datenkategorien gemäß der ISO-Norm 12620-1 (2022) und DatCatInfo aus. Die Systemgebundenheit des Rechts kann für die Verwaltung der Datenkate‐ gorie / Definition/ auf Begriffs- oder auf Sprachebene entscheidend sein. Damit verbunden ist die Zielsetzung der jeweiligen Terminologiesammlung, die entwe‐ der rechtssystemübergreifend und rechtsvergleichend oder rechtssystemintern sowie ein- oder mehrsprachig angelegt werden kann. Hinzu kommen Entschei‐ dungen über einzelne Datenkategorien, die für die technische Kommunikation, Translation, Wissensorganisation oder Sprachplanung besonders relevant sind. Schließlich sind die Erwartungen und Anforderungen der Nutzergruppe(n) zu berücksichtigen. Dazu gehören Rechtsfachleute, Rechtstranslator: innen sowie die allgemeine Bevölkerung. Zunehmend sind auch Maschinen als mögliche Nutzerinnen zu berücksichtigen. Für Letztere sind international anerkannte und maschinenlesbare Formate besonders wichtig. Eine zentrale Aufgabe bei der Konzeption und Strukturierung von Termino‐ logiedatenbanken besteht darin, unter Beachtung der terminologischen Grund‐ sätze einen tragfähigen Kompromiss zwischen den verschiedenen Einflussfak‐ toren zu finden. Diese hängen teilweise voneinander ab und sollten gemeinsam berücksichtigt werden. Welche Anforderungen dabei jeweils im Vordergrund stehen und inwieweit etwaige Kompromisslösungen sinnvoll sind, muss von Fall zu Fall entschieden werden. Terminologiearbeit erfolgt zweck- und nutzer‐ orientiert, wobei mit der Zeit auch Änderungen bestimmter Anforderungen eintreten könnten. Allerdings ist eine Überarbeitung des Datenmodells und der Datenstruktur nicht immer möglich oder kann mit beachtlichem Aufwand verbunden sein. Daher ist es immer empfehlenswert, von Anfang an möglich weitsichtig zu planen. Bibliographie Arntz, Reiner/ Picht, Heribert/ Schmitz, Klaus-Dirk (2021). Einführung in die Terminolo‐ giearbeit. 8. Aufl. Hildesheim: Olms. Bestué, Carmen (2019). A Matter of Justice. Integrating Comparative Law Methods into the Decision-Making Process in Legal Translation. In: Biel, Łucja/ Engberg, Jan/ Ruano, Rosario Martín/ Sosoni, Vilelmini (Hrsg.). Research Methods in Legal Translation and Interpreting. Crossing Methodological Boundaries. London: Routledge, 130-47. Biel, Łucja (2014). Phraseology in Legal Translation. In: Cheng, Le/ Kui Sin, King/ Wagner, Anne (Hrsg.). The Ashgate Handbook of Legal Translation. Farnham: Ashgate, 177-92. 116 Elena Chiocchetti, Natascia Ralli <?page no="117"?> bistro (o. J.). Informationssystem für Rechtsterminologie. Aufrufbar unter: https: / / bistro .eurac.edu/ de/ (Stand: 28.04.2025) Budin, Gerhard (2019). Zum Entwicklungsstand der Terminologiewissenschaft. In: Dre‐ wer, Petra/ Pulitano, Donatella (Hrsg.). Terminologie: Epochen - Schwerpunkte - Umsetzungen. Zum 25-jährigen Bestehen des Rats für Deutschsprachige Terminolo‐ gie. Berlin: Springer Vieweg, 7-19. Cao, Deborah (2007). Translating Law. Clevedon: Multilingual Matters. Chiarcos, Christian/ McCrae, John P. / Cimiano, Philipp/ Fellbaum, Christiane (2013). Towards Open Data for Linguistics: Linguistic Linked Data. In: Oltramari, Alessan‐ dro/ Vossen, Piek/ Hovy, Eduard (Hrsg.). New Trends of Research in Ontologies and Lexical Resources. Ideas, Projects, Systems. Berlin/ Heidelberg: Springer, 7-25. Chiocchetti, Elena/ Lušicky, Vesna/ Wissik, Tanja (2023). Multilingual Legal Terminology Databases: Workflows and Roles. In: Biel, Łucja/ Kockaert, Hendrik J. (Hrsg.). Hand‐ book of Terminology, vol. 3: Legal Terminology. Amsterdam/ Philadelphia: John Benjamins, 458-484. Cimiano, Philipp/ John P. McCrae/ Victor Rodríguez-Doncel/ Gornostay, Tatiana/ Gó‐ mez-Pérez, Asunción/ Siemoneit, Benjamin/ Lagzdiņš, Andis (2015). Linked Termino‐ logies: Applying Linked Data Principles to Terminological Resources. In: Kosem, Iztok/ Jakubíček, Miloš/ Kallas, Jelena/ Krek, Simon (Hrsg.). Proceedings of the eLex 2015 Conference. Ljubljana/ Brighton: Trojina/ Lexical, 504-17. Danzl, Karl-Heinz (2024). § 6 EKHG. Abrufbar unter: rdb.manz.at/ docu‐ ment/ 1222_2_ekhg_p0006 (Stand: 20.01.2026) DatCatInfo (o. J.). Data Category Repository. Abrufbar unter: https: / / datcatinfo.termwe b.eu/ client/ twd (Stand: 28.04.2025) de Groot, Gerard-René (1999). Das Übersetzen juristischer Terminologie. In: de Groot, Gerard-René/ Schulze, Rainer (Hrsg.). Recht und Übersetzen. Baden Baden: Nomos, 11-46. Di Buono, Maria Pia/ Cimiano, Philipp/ Elahi, Mohammad Fazleh/ Grimm, Frank (2020). Terme-à-LLOD: Simplifying the Conversion and Hosting of Terminological Resources as Linked Data. In: Ionov, Maxim/ McCrae, John P./ Chiarcos, Christian/ Declerck, Thierry/ Bosque-Gil, Julia/ Gracia, Jorge (Hrsg.). Proceedings of the 7th Workshop on Linked Data in Linguistics (LDL-2020). Paris: European Language Resources Association, 28-35. DIN ISO 20539 (2024). Übersetzen, Dolmetschen und zugehörige Technik - Begriffe. Berlin: Beuth. DIN SPEC 91526 (2025). Knowledge Graphs for Language Models and Language Models for Knowledge Graphs - Hybrid Applications of symbolic and subsymbolic AI. Berlin: Beuth. Einflussfaktoren bei der Terminologiearbeit im Recht 117 <?page no="118"?> Drewer, Petra/ Schmitz, Klaus-Dirk (2017). Terminologiemanagement. Grundlagen - Methoden - Werkzeuge. Berlin: Springer Vieweg. Durán Muñoz, Isabel (2012). La ontoterminografía aplicada a la traducción. Propuesta metodológica para la elaboración de recursos terminológicos dirigidos a traductores. Frankfurt: Peter Lang. EIGE - European Institute for Gender Equality. (o. J.) Legal Definitions in the EU Member States. Abrufbar unter: https: / / eige.europa.eu/ gender-based-violence/ regulatory-and -legal-framework/ legal-definitions-in-the-eu (Stand: 28.04.2025) Gambaro, Antonio/ Sacco, Rodolfo (2018). Sistemi giuridici comparati. 4. Aufl. Milano: UTET. Gesetz 76 (2016). Legge 20 maggio 2016, n. 76. Regolamentazione delle unioni civili tra persone dello stesso sesso e disciplina delle convivenze. Abrufbar unter: http: / / www .gazzettaufficiale.it/ eli/ id/ 2016/ 05/ 21/ 16G00082/ sg (Stand: 28.05.2025) Gesetzesvertretendes Dekret 154 (2013). Decreto legislativo 28 dicembre 2013, n. 154. Revisione delle disposizioni vigenti in materia di filiazione, a norma dell’articolo 2 della legge 10 dicembre 2012, n. 219. Abrufbar unter: http: / / www.gazzettaufficiale.it/ eli/ id/ 2014/ 01/ 08/ 14G00001/ sg (Stand: 28.05.2025) ISO 639 (2023). Code for Individual Languages and Language Groups. Genf: International Organization for Standardization. ISO 1087 (2019). Terminology Work and Terminology Science — Vocabulary. Genf: International Organization for Standardization. ISO 3166-1 (2020). Codes for the Representation of Names of Countries and Their Sub‐ divisions. Part 1: Country Code. Genf: International Organization for Standardization. ISO 3166-2 (2020). Codes for the Representation of Names of Countries and Their Subdivisions. Part 2: Country Subdivision Code. Genf: International Organization for Standardization. ISO 12620-1 (2022). Management of terminology resources — Data categories — Part 1: Specifications. Genf: International Organization for Standardization. ISO 26162-1 (2019). Management of Terminology Resources - Terminology Databases - Part 1: Design. Genf: International Organization for Standardization. ISO 30042 (2019). Management of Terminology Resources - TermBase eXchange (TBX). Genf: International Organization for Standardization. ISO/ TS 24634 (2024). Management of Terminology Resources - TermBase eXchange (TBX)-compliant representation of concept relations and subject fields. Genf: Inter‐ national Organization for Standardization. Martín-Chozas, Patricia (2019). Creation and Enrichment of a Terminological Knowledge Graph in the Legal Domain. In: Palmirani, Monica (Hrsg.). Proceedings of the Seventh JURIX 2019 Doctoral Consortium. Madrid: CEUR Workshop Proceedings Vol 2598. 118 Elena Chiocchetti, Natascia Ralli <?page no="119"?> Montiel-Ponsoda, Elena/ Bosque-Gil, Julia/ Aguado de Cea, Guadalupe/ Vila-Suero, Daniel (2015). Towards the Integration of Multilingual Terminologies: An Example of a Linked Data Prototype. In: Poibeau, Thierry/ Faber, Pamela (Hrsg.). Proceedings of the 11th International Conference on Terminology and Artificial Intelligence. Granada: CEUR Workshop Proceedings, Vol. 1495, 205-206. Muhr, Rudolf/ Peinhopf, Marlene (2015). Wörterbuch rechtsterminologischer Unter‐ schiede Österreich-Deutschland. Frankfurt am Main: Peter Lang. Nielsen, Sandro (2023). Legal Lexicography and Legal Information Tools. In: Biel, Łucja/ Kockaert, Hendrik J. (Hrsg.). Handbook of Terminology. Legal Terminology, 3. Amsterdam/ Philadelphia: John Benjamins, 432-457. Peruzzo, Katia (2018). Risorse terminologiche su misura: indagine e riflessione sulle necessità e le preferenze degli studenti di Giurisprudenza. In: Magris, Marella (Hrsg.). La banca dati TERMitLEX: un nuovo modello interdisciplinare per la terminografia giuridica. Trieste: EUT, 87-106. RaDT - Rat für Deutschsprachige Terminologie (2015). Terminologie. Grundlage für Fachkommunikation, Fachinformation, Fachwissen. Rat für Deutschsprachige Termi‐ nologie. Abrufbar unter: https: / / publikationen.radt.org/ RaDT_Terminologie_2015_d eutsch.pdf (Stand: 28.05.2025) Ralli, Natascia (2025). Managing Language Varieties: Examples From Legal Terminology Work. In: Vezzani, Federica/ Di Nunzio, Giorgio Maria/ Loupaki, Elpida/ Meditskos, Georgios/ Papoutsoglou, Maria (Hrsg.). Proceedings of the 4th International Confe‐ rence on Multilingual Digital Terminology Today. Design, Representation Formats and Management Systems (MDTT 2025). Thessaloniki: CEUR Workshop Proceedings, Vol. 3990. Ralli, Natascia/ Andreatta, Norbert (2018). bistro - ein Tool für mehrsprachige Rechtster‐ minologie. trans-kom 11 (1), 7-44. Rodríguez-Doncel, Victor/ Santos, Cristiana/ Casanovas, Pompeu/ Gómez-Pérez, Asun‐ ción/ Gracia, Jorge (2018). A Linked Data Terminology for Copyright Based on Ontolex-Lemon. In: Rodríguez-Doncel, Víctor/ Palmirani, Monica/ Araszkiewicz, Mi‐ chał/ Casanovas, Pompeu/ Pagallo, Ugo/ Sartor, Giovanni (Hrsg.). AI Approaches to the Complexity of Legal Systems. Berlin: Springer, 410-423. Sandrini, Peter (1996a). Terminologiearbeit im Recht. Deskriptiver begriffsorientierter Ansatz vom Standpunkt des Übersetzers. Wien: Braumüller. Sandrini, Peter (1996b). Comparative analysis of legal terms: equivalence revisited. In: Galinski, Christian (Hrsg.). Terminology and knowledge engineering. Frankfurt am Main: Indeks, 342-350. Sandrini, Peter (2014). Multinational Legal Terminology in a Paper Dictionary. In: Mac Aodha, Máirtín (Hrsg.). Legal Lexicography. A Comparative Perspective. London: Routledge, 141-152. Einflussfaktoren bei der Terminologiearbeit im Recht 119 <?page no="120"?> Sandrini, Peter (2018). Terminology Work in Different Domains: Legal Terminology. In: Humbley, John/ Budin, Gerhard/ Laurén, Christer (Hrsg.). Languages for Special Purposes: An International Handbook. Berlin/ Boston: De Gruyter Mouton, 509-521. Šarčević, Susan (1997). New Approach to Legal Translation. The Hague: Kluwer Law International. Siewer-Kowalkowska, Katarzyna (2017). Zum lexikalischen und semantischen Sprach‐ wandel der deutschen und polnischen Rechtssprache am Beispiel des strafrechtlichen Wortschatzes. Comparative Legilinguistics 29: 50-65. Tarp, Sven (2017). The Third Leg of Two-legged Lexicography. Hermes 21 (40): 117-31. Vanden Bulcke, Patricia/ De Groote, Carina (2016). JuriGenT, un banco de datos jurídico neerlandés/ español diferente. CLINA 2 (2): 15-38. Vezzani, Federica/ Di Nunzio, Giorgio Maria/ Henrot, Geneviève (2018). TriMED: A Multilingual Terminological Database. In: Calzolari, Nicoletta/ Choukri, Khalid/ Dec‐ lerck, Thierry/ Lopez, Adriana/ Maziero, Rosita/ Monachini, Massimiliano/ Ono, Nao/ Pezzelle, Samuele (Hrsg.). Proceedings of the 11th Edition of the Language Resources and Evaluation Conference. Miyazaki: ACL Anthology, 4367-4371. VfGH - Erkenntnis des Verfassungsgerichtshofs vom 04.12.2017, G 258-259/ 2017-9. Abrufbar unter: https: / / www.vfgh.gv.at/ downloads/ VfGH_Entscheidung_G_258-201 7_ua_Ehe_gleichgeschlechtl_Paare.pdf (Stand: 28.05.2025) Warburton, Kara/ Wright, Sue Ellen (2019). A Data Category Repository for Language Resources. In: Pareja-Lora, Antonio/ Alonso-Ramos, Manuel/ Aguado de Cea, Guada‐ lupe (Hrsg.). Development of Linguistic Linked Open Data Resources for Collaborative Data-Intensive Research in the Language Sciences. Cambridge, Mass./ London: Mas‐ sachusetts Institute of Technology, 69-97. Wetzel, Michael/ Mathioulakis, Stelios/ Hummel, Jochen/ Vasilevich, Alena (2021). Colla‐ borative-AI Knowledge Graph Generation: Taxonomization of IATE, the EU Termino‐ logy. In: Chaves-Fraga, David/ Dimou, Anastasia/ Heyvaert, Pieter/ Priyatna, Freddy/ Sequeda, Juan (Hrsg.). Proceedings of the Second International Workshop on Know‐ ledge Graph Construction (KGCW 2021). Hersonissos: CEUR Workshop Proceedings, Vol. 2873. 120 Elena Chiocchetti, Natascia Ralli <?page no="121"?> Terminology at the Dutch Language Institute: Towards a centre of expertise for professional language Frieda Steurs Abstract: In my career, I had the pleasure of working with Gerhard Budin for many years. Being both professors in the field of terminology, teaching at university, supervising PhD students, working in the ISO TC/ 37 committee and cooperating in TermNet, the international terminology network. I shifted careers some eight years ago, and became the head of research of the Dutch Language Institute in Leiden. I was now able to create and build a center of expertise for Dutch terminology. In my paper, I will present the different types of information that is made available by the ENT. This also includes specific software tools for terminology work. Finally, some of our research projects on terminology will be touched upon. 1 Introduction The Dutch language area has a longstanding tradition in cooperation in the field of terminology. Since the eighties, different partnerships and commissions existed, all under the umbrella of the Union of the Dutch language (Taalunie). A cooperation agreement existed on the topic “Terminology and Education”, there was a commission for terminology (Coterm) that was responsible for the strate‐ gic planning for Dutch terminology, and an association for Dutch Terminology, NL-Term was founded. Gradually, the idea developed that a central information hub was needed to bring together all the information on Dutch language for professional use and to make this available to everyone, both specialists and the general public. Since 2016, The Dutch Language Institute in Leiden (INT) is responsible for the development and hosting of all language material for Dutch. Whereas formerly, only the standard language was researched and collected, the new linguistic paradigm includes all language varieties and as such also professional language, regional languages and dialects. It seemed a logical step <?page no="122"?> to also host the Center of Expertise for Dutch Terminology (ENT) at the INT. The main task of the ENT is to provide all kinds of information, products and services related to terminology. Next to making all the information available, ENT also works on a number of research projects in specific domains: legal and medical language and the terminology used in higher education. 2 Terminology in the Dutch-speaking regions: A short history The Dutch-speaking regions have a long tradition of (linguistic) collaboration. Since the eighties there have been several partnerships and committees, all lead by the Dutch Language Union (NTU). Linguistic ideas saw significant change in the eighties. In the twentieth century, theoretical linguistics had evolved from structuralism to generative grammar, focusing on syntax. The approach to the lexicon was still rather traditional, with an emphasis on standard language, modern and historical. Because of growing internationalisation and the consequential increased demand for translators and interpreters capable of dealing with specialist texts, professional language received more and more attention, although not immediately from the side of traditional, theoretical linguistics. The NTU realised there was a need for a consultation platform on the subject of professional language. There were two other interested parties: the Royal Netherlands Standardization Institute (then NNI, now NEN) and the Belgian Bureau for Standardisation (then BIN, now NBN). A partnership was started, which is the Dutch terminology group Samenwerkingsverband voor Nederlandse Terminologie (SaNT). The SaNT was working closely together in the period 1988-1996. Within this collaboration, interested parties were brought together in two groups: a Language and Education group and a Language and Business group. In the series Voorzetten (‘proposals’) by the Dutch Language Union, two re‐ ports on terminology were published. The first report (Vroomen 1990) included articles on the promotion of the use of authorised terminology lists, on how to use terminology databases in businesses and government institutions, and also an article on how universities should offer Terminology as a subject in its own right (Vervoorn 1992). Recommendations were written on the contents of such courses and on how research into professional language could be given shape. As a result of these recommendations by the SaNT, all translation and interpretation programmes in Flanders and the Netherlands began to pay more attention to terminology, making it a separate subject in the curriculum. At the time, this was highly innovative, as most language and translation courses still maintained an overwhelmingly philological approach. 122 Frieda Steurs <?page no="123"?> 1 https: / / cordis.europa.eu/ project/ id/ LRE63090/ results This was also the time of the POINTER project. POINTER stood for Proposals for an Operational Infrastructure for Terminology in Europe and was launched in December 1994 by the European Commission. One of the results of this project was the foundation of the European Association for Terminology (EAFT). Even then there was an awareness that professional language was becoming very important in European businesses and that a multilingual policy was required. Many companies and organisations use and invent terms on a daily basis, not only in their own institutions and countries, but also across regional and national borders. Although new and frequently occurring terminology is often recorded, there is a lack of rationalisation despite our so-called data-driven information society and hardly any efforts are made to work together to maximise the use of valuable sources 1 . The POINTER project aimed to make a set of concrete, practical proposals in order to support terminology users all over Europe by facilitating the distribution of terminologies. The emphasis of POINTER was on meeting the multilingual needs of a growing European market. These were fascinating times: most researchers had gradually switched over from paper to working digitally and were learning to use their first PCs. The importance of digital linguistic data to researchers and users became clear. The SaNT working group ‘Language and Business’ experimented with the first digital termbase: Eurotermbank (Maastricht) and Eurodicautom (European institutions, Luxembourg, later replaced by IATE). In all activities of these two working groups, we noticed an increasing interest in professional language. Economic activity and the use of professional language are of course closely interwoven. Because of the strong economic markets in the Netherlands and Flanders, many businesses needed professional texts for customers, suppliers and technicians. These texts had to be quickly available in different languages, which caused a need for terminology lists and later termbases. This phenomenon has certainly contributed to the development of a consultation platform for Dutch terminology. The activities of the SaNT came to an end in 1996, when this collaboration was discontinued by the NTU. As the standardisation institutes had hardly contributed to this initiative, the NTU did not consider it a real partnership between the parties mentioned. The colleagues concerned, terminologists, linguists and translators who had been working together with the SaNT, were left on their own and decided in 1997, on the occasion of a congress in Utrecht, to found a field association for Dutch terminology for the Netherlands and Flanders. Thus NLTerm was born, an association which aimed to be a contact Terminology at the Dutch Language Institute 123 <?page no="124"?> 2 https: / / www.nlterm.eu/ 3 https: / / iate.europa.eu/ point for anyone involved with Dutch terminology, both its practice and its theory. In its first years NLTerm was kept afloat by volunteers and enthusiastic members of the association, without any external funding. Later a subsidy was granted by the NTU. NLTerm has been successful now for over 30 years and organises a congress or a workshop every year 2 . After cancelling the SaNT, the NTU did not abandon terminology, but appointed a strategic committee which was to establish the policy and priorities for terminology in the Dutch-speaking regions. CoTerm was founded in 2006 and drew up policy decisions until 2012. It was the committee’s task to advise the NTU on the terminology policy to be implemented within the long-term policy framework of the NTU, to formulate projects within that framework and to carry them out or have them carried out under its own supervision. Within the committee, the exchange of expertise was promoted. Eurodicautom was the pioneering terminology database of the European Commission, set up in 1975, at first, to be used by translators and other em‐ ployees of the Commission. By 1980, it was accessible online to the Commission. During the eighties and nineties, the database grew, but accessing it from outside of the European institutions still required dialling in via a modem. At first only Dutch, French, German and Italian were represented, the four languages of the six countries that formed the EEC (the three Benelux countries, France, Germany and Italy). When more countries joined the European Union, other languages were added, the number increased to six, then seven, nine and eventually eleven languages (Latin was added for scientific names). Later, public user interfaces were added, giving the larger public free access to multilingual terminology related to European Union activities. The Ministries of Foreign Affairs of several member states added terminology to the database. Senior terminologists worked on the evaluation and creation of new entries in the existing database in different languages. In 2007, Eurodicautom was replaced by Interactive Terminology for Europe (IATE) 3 . On June 28th, 2007, the European institutions sent out a press release to announce that three existing termbases were merged into a single EU terminology database: IATE, which covered 23 languages and was open to the public. There were two particular features distinguish IATE from its predecessors Eurodicautom (European Commission), Euterpe (European Parliament) and TIS (Council of the European Union). First, IATE is interactive; any translator in an EU institution can add and update information in the database. To ensure the quality of the individual contributions, a change in 124 Frieda Steurs <?page no="125"?> 4 http: / / www.infoterm.info/ 5 https: / / www.termnet.org/ the database automatically launches a validation cycle: terminologists of the translation departments validate the new and modified information. Secondly, IATE is interinstitutional; its project partners are the European Parliament, the Council, the Commission, the Court of Justice, the Central Bank, the Court of Auditors, Economic and Social Committee, the Committee of the Regions, the European Investment Bank and the Translation Centre. However, not all knowledge and insights regarding terminology came from the translation community. As it happens, the first steps in the theorization of terminology were made by engineers and professional experts and were strictly monolingual. It all started with the ‘Viennese school’, which laid the foundations of classical terminology, articulated by Eugen Wüster (1898-1977). Wüster was an Austrian engineer and the founder of Terminology as a discipline. He got a degree in electrical engineering but was above all fascinated by the science of terminology. Living in a time when technology was developing rapidly, he asked himself questions about the way in which new concepts are reflected in language. His Dissertation on the Internationale Sprachnormung in der Technik, besonders in der Elektrotechnik (International language standardisation in technology, particularly in electrical engineering) (Wüster 1931) was regarded as a terminological standard work. He taught at the University of Vienna, where af‐ ter his death the Institut für Wissenschaftstheorie und Wissenschaftsforschung and the Eugen Wüster Archive were founded. His research into technical communication resulted in the establishment of the Technical Committee for Standardization of the International Organization for Standardization (ISO/ TC 37). Wüster developed the General Theory of Terminology (GTT) (Wüster 1979). He was involved in Infoterm 4 , the International Information Centre for Terminology founded in 1971 by UNESCO. Infoterm aims to support and coordinate international collaboration in the field of terminology. The members of Infoterm are national, international and regional institutes, organizations and private specialists, in short, everyone who has anything to do with the standardization and harmonization of professional language. It is a worldwide platform, with members on all continents. Infoterm and the later founded TermNet 5 , the international network for terminology, will keep playing an important role in the dissemination of terminological insights, the organization of workshops and the promotion of terminology applications in businesses and governments. Terminology at the Dutch Language Institute 125 <?page no="126"?> 6 https: / / www.tekom.de/ 7 https: / / www.technical-communication.org/ Another important player in the international field is Tekom. Both Tekom Germany 6 and Tekom Europe 7 are very large professional associations for tech‐ nical communication. Their activities are worldwide and of great importance in the world of technical and professional documentation. Again, terminology is one of their core tasks. Following their example, we were successful in creating a Tekom Belgium. Later on, a Tekom Netherlands was also founded. In the following sections we will specifically look into the situation regarding Dutch, especially with regard to how all data concerning terminology can be centralized and promoted. 3 The Dutch terminology institutions and projects In 2016, the Institute for Dutch Lexicology transformed into the more broadly oriented Dutch Language Institute. This change went hand in hand with the renewed task of the General Secretariat of the NTU, which was to focus on policy tasks, leaving the executive tasks to the INT. To the Dutch Language Institute this transfer of tasks offered the opportunity to broaden its own activities. The institute became the central point of contact regarding the keeping and maintenance of digital language materials and the safekeeping of data collections related to any variations of Dutch. This evolution reflects the strongly altered landscape of linguistic research: large language infrastructures are digitally set up and contain the corpora, dictionaries and other specialised lexicons and databases, grammar and so much more. The institute develops and provides data for dictionaries, (computational) lexicons, corpora and tools. The dictionaries are accessible online. Software and computational linguistic tools are available as open source. The INT takes up a central position in the whole of the Dutch-speaking world (the Netherlands, Flanders, Suriname and the Caribbean) as a developer, keeper and distributor of scholarly and sustainable language resources. The institute is well-equipped for this task and has a large international network for the exchange of information with like-minded institutions. The Dutch Language Institute also provides the necessary building blocks for all language applications aimed at the development and improvement of businesses and public organisations. We intend to strengthen this role in the coming years, which is why we are focusing on the sustainable distribution of any language materials, with an emphasis on: 126 Frieda Steurs <?page no="127"?> 8 https: / / www.perlego.com/ knowledge/ study-guides/ what-are-the-digital-humanities/ 9 https: / / www.clarin.eu/ - the Dutch vocabulary, both historical and contemporary, both standard language and the dialects, both general language and professional language; - new technologies and techniques to make the internet accessible for lin‐ guistic research and for the ongoing maintenance of constantly updated, extensive corpora of contemporary Dutch; - contribution to the accessibility of historical text material (coming from inside and outside of the INT), in which the considerable spelling variation is no longer a search impediment, and ways are offered to detect and circumvent variations in word use; - use of and contribution to new computational linguistic or language tech‐ nology techniques in aid of information retrieval from language materials; - formal structuring of linguistic information, making it suitable for compu‐ tational linguistic applications; - realisation of facilities for third parties to contribute interactively to the description of the Dutch language and the optimization of the central digital data infrastructure for this purpose; - becoming a point of contact for all language teachers and building an infrastructure of language materials that are useful and necessary for the support of teaching Dutch to various types of language learners. The institute responds to the new developments in the field of Digital Hu‐ manities 8 . In order to fulfil this role, the INT keeps and maintains a digital infrastructure for the Dutch language, paying attention to language variation (terminology, dialects, etc.). Both academic and non-academic parties can make use of this infrastructure. The INT sees a clear overlap between its own activities - the central data infrastructure - and recent developments within the eHumanities. With its own expertise, the INT contributes to the digital future of the humanities in the Netherlands and Flanders. On the one hand, knowledge and products are delivered in aid of other scientific organisations, and on the other hand, collaboration with the eHumanities enhances the quality of the central data infrastructure of the Dutch language. We work closely together with centres for digital humanities at various universities and with networks such as the KNAW Humanities Cluster (Netherlands), Digital Humanities Benelux and the WOG Digital Humanities (Flanders). The INT functions as a CLARIN centre for Flanders and informs Flemish researchers about the latest developments in the field of linguistic sources and the wider linguistic infrastructure in Europe (CLARIN ERIC) 9 . This allows every researcher to learn more about access to Terminology at the Dutch Language Institute 127 <?page no="128"?> repositories, standards, metadata, available corpora, the methods to encode their own corpus material, storage facilities, etc. That an infrastructure like the INT’s can become a real digital platform of all language materials is an interesting and quite recent development. For a long time, attention both in theoretical and in applied linguistics had been focused on the General Standard Language (LGP), which was studied from a diachronic and a synchronic point of view. Public institutions for language were not expected to build collections of dialects, regiolects or specialist professional language (LSP). Only when the insights into language variation changed, the step was made to a wider research field, describing the complete language spectrum with attention to all registers (see Fig. 1). From highly specialist terms to more common terms for technical or scientific concepts in standard language, to local speech, regional variation and informal usage: the whole range is now covered. It was no more than a logical step to embed the Centre of Expertise for Dutch Terminology (ENT) in an existing institute that makes all language materials available. Fig. 1: The connection between general language and professional language: a conti‐ nuous spectrum 128 Frieda Steurs <?page no="129"?> 10 https: / / ivdnt.org/ terminologie/ expertisecentrum/ The ENT is a junction in a wider network of relations both with the terminolo‐ gical field and with more general social sectors and wider user groups. Two lines of policy are being further developed: - The ENT provides diversified supporting facilities regarding terminology, especially Dutch terminology. - The ENT positions itself in the terminological network by means of contact or collaboration with terminological associations or terminology-related institutions, both nationally and internationally. The ENT is part of the wider INT organisation. For the sake of the recognisability of this structural connection and also to make use of the wide reputation of the INT website, the ENT has not developed a separate terminology website but uses a varied terminology page within the INT website. This terminology page contains the following features/ sections, regularly kept up to date by the ENT. Fig. 2: Website of the Centre of Expertise of Dutch terminology On the website 10 (see Fig. 2), users will find a short introduction on terminology and an extensive overview of relevant sources. For those who would like to gain more in-depth knowledge, the ENT provides a bibliographical list with profes‐ sional literature (manuals and theories/ models) in the field of terminology. One of the key tasks of the ENT is to make terminological collections available. The internet hosts rich terminological collections belonging to an Terminology at the Dutch Language Institute 129 <?page no="130"?> 11 https: / / onzetaal.nl/ schatkamer/ kijken-en-luisteren/ podcasts/ frieda-steurs 12 https: / / termwerk.ato.ivdnt.org/ enormous range of professional languages. Both for the benefit of the profes‐ sional field of terminologists, translators etc. and for interested people from the wider public, monolingual Dutch lists and multilingual lists featuring Dutch are being collected. These terminology collections are gradually extended and provided, along with hyperlinks. The option for users to put forward new terminological collections is in the pipeline. The ENT terminology page should become the place for terminology database and dictionaries to be presented and made available with hyperlinks. To keep this collection organised and manageable, the term collections are divided according to the internationally used Library of Congress classification system. In principle, the wide angle of the ENT accommodates all professions. Terminological collections can be the result of finished projects but can also be subject to maintenance, updates or extensions. The Centre of Expertise also maintains an overview of the software tools available for terminology work. It concerns tools that are meant for tasks such as aligning, corpus analysis, online concordances, term extraction, terminology control, translation memories with integrated terminology module, and search applications. The ENT provides newsletters, information on terminology courses and an interactive map for the whole of Europe with terminological institutions. To help disseminate knowledge about terminology, we created a podcast 11 and a webinar on terminology tools. 4 Term extraction and terminology management: the development of tools In cooperation with the Dutch Language Union, the ENT supports the termi‐ nology field with tools and software. In 2024, the INT developed the new tool TermWerk 12 , a web-based application for automatic terminology extraction and management of terminology databases. This online working environment integrates the functions of two older tools, the TermTreffer and the TermBe‐ heerder, previously developed on behalf of the Dutch Language Union and made available by the INT for several years. The replacement of these externally developed tools had become necessary because the software proved difficult to maintain, so INT could not guarantee its long-term availability. In contrast, the new environment is fully integrated into INT’s other software systems. Based 130 Frieda Steurs <?page no="131"?> on recent scientific insights, new terminology components were implemented, such as the automatic terminology extraction module, a database structure and a TBX export function that comply with the latest ISO standards. All in all, embedding TermWerk in the INT systems ensures sustainable availability, support and further development of the application. The implementation offered the opportunity to bring together various func‐ tions in one online working environment. Whereas users previously had to first save the results of an automatic term extraction and then upload them again to a term management application, this can now be done within the same application. Within a TermWerk project, users can work together to build and publish a term bank in four steps: 1. Corpus creation and corpus analysis: Users can upload Dutch-language subject texts, linguistically enrich them and then search them in the corpus search environment for words, expressions or morphosyntactic patterns. Within one project, several corpora can be created and provided with metadata in order, for example, to be able to handle several subdomains or language varieties in the terms database at a later stage. 2. Term extraction: from the created corpora, users can automatically extract lists of both one-word and multi-word terms ordered by their domain specificity (termhood). Term lists can be compared with each other and possibly checked against an exclusion list. Terms can be given a first edit, e.g. lemmatization correction, and then selected to be transferred to the term database for a more comprehensive terminological description. The application also keeps track of which terms are already in the term bank. 3. Term management: The project owner can create a term bank, determine which languages are included and set the desired (TBX-compliant) desc‐ ription fields for each description level (concept/ language/ term). As in other terminology management systems, project staff can then supplement the term bank with term records and information at concept, language or term level (domain, definition, contexts, etc.) and a record is kept of which editing stage each term record is in. However, what makes TermWerk special is that, in addition, selected terms can be transferred to the term database from a term list. The user can choose whether the terms are imported as ‘floating’ term records that will be linked to a concept at a later stage, or whether a new concept is immediately created for each transferred term, which is then further processed in the term database. 4. Online publication: A term bank created in TermWerk can not only be exported as a TermBase eXchange (TBX) file but can also be published online immediately on the TermWerk website, making it searchable for Terminology at the Dutch Language Institute 131 <?page no="132"?> everyone. However, TermWerk users remain themselves responsible for the published content. The development of the TermWerk application was completed in 2024 and a first version was tested in January 2025. During 2025, TermWerk will be made available to the general public free of charge. Anyone with an email account from a European college, university or research institute will then be able to log in directly via the CLARIN single sign on. Those without an academic account will be able to apply for an account through the TermWerk website. 5 Termraad for the Dutch Language The Termraad for Dutch was set up in 2013, when the need for it was felt by the EU institutions. It is an informal consultation platform for terminological conformity within the Dutch language regions. The Termraad Nederlands is part of a terminological partnership between the Dutch Language Council and the translation services of the EU and the Dutch, Belgian and Flemish authorities. The partners consult each other to find uniform translations for EU-terminology in Dutch. The following institutions are represented in the core group of the Termraad Nederlands: - European Commission - Council of the European Union - European Parliament - European Court of Auditors - Court of Justice of the European Union - Dutch government - Belgian government - Flemish government - Dutch Language Institute (INT) - Dutch Language Union (NTU) Within the Termraad there is regular consultation about terms such as those used in higher education, in order to establish a uniform translation. Also, other terms that are relevant to IATE and that are related to European policy are discussed. EU translators are expected to use language and terminology that are member state-neutral. Any ambiguity, inconsistency and doubt are to be avoided. A consistent terminology means that the same terms are always used to denote the same concepts. Dutch is, just like English, French and German, a pluricentric language. Because the Netherlands and Belgium/ Flanders are dif‐ 132 Frieda Steurs <?page no="133"?> 13 https: / / ec.europa.eu/ info/ resources-partners/ european-masters-translation-emt_en ferent societies with different forms of administration, legal and governmental traditions, and cultural and social attitudes, it is inevitable that the national varieties should reflect these differences. After the successful start of the Termraad, the Termraad Academy was set up to bridge the gap between the Directorate-General for Translation and final-year students, specifically for Belgian and Dutch students of the EMT (European Master in Translation) 13 courses in the translation master. The Termraad Aca‐ demy aims to offer students a suitable internship or an interesting subject for a Master’s degree thesis. It is concerned with the collection, analysis and distribution of field-specific terminology. For Dutch universities, the INT/ ENT is functioning as a host to guide these students and to show them the way in the world of terminology. Partners are the Directorate-General for Translation of the European Com‐ mission, the Council of the EU and the Benelux Union. The Termraad Academy can help bridge the gap between Flanders and the Netherlands (and the EU) and improve collaboration between higher education and the labour market. At the same time, the IATE terminology database for Dutch profits from the many valuable additions that the project produces. 6 Conclusion The creation of a Centre of Expertise for Dutch Terminology has allowed us to bring together a lot of interesting information on professional language and on how the Dutch language has developed in the field of science and technology. The centre also makes it possible to build a large network together with governments, businesses, international institutions and users such as teachers and translators. Such a centre can certainly be of great extra value and maybe serve as a source of inspiration for other language communities. Bibliography Vervoorn, Ad (1992). Samenwerkingsverband Nederlandstalige Terminologie. Termino‐ logieleer: Beleid en Praktijk. In: Voorzetten 38. The Hague: Stichting Bibliographia Neerlandica. Vroomen, Pim de (1990). Terminologie in het Nederlandse Taalgebied. Verslagen en rapporten over het eerste driejarenplan (1988-2000) van het Samenwerkingsverband Terminology at the Dutch Language Institute 133 <?page no="134"?> Nederlandstalige Terminologie (SaNT). In: Voorzetten 32. The Hag: Stichting Biblio‐ graphia Neerlandica. Wüster, Eugen (1931). Internationale Sprachnormung in der Technik, besonders in der Elektrotechnik. Wien: Technische Hochschule Wien. Dissertation. 134 Frieda Steurs <?page no="135"?> A Constructionist Pictorial Approach Towards a Universal Language Representation Werner Winiwarter ちりも積もれば山となる Even dust, when piled up, can become a mountain. J A P A N E S E P R O V E R B Abstract: We present a novel language representation based on five pillars: uniquely identifiable concepts, visualization, Radical Construction Grammar (RCG), Uniform Meaning Representation (UMR), and Interlinear Morphemic Glossing (IMG). The resulting annotation is well suited for multilingual applications, especially for second language acquisition. We have integrated the representation into a web-based language learning environment to allow instructors to design engaging annotation exercises using augmented browsing technology. In our use cases, we focus on the Japanese language because it poses unique challenges for European learners. We also discuss extensions towards translation deviation anno‐ tations, support for under-resourced languages, and various user interface design choices based on lessons learned in augmentative and alternative communication, colourful semantics, and iconic and international auxili‐ ary languages 1 Introduction We present a novel language representation based on five pillars: uniquely iden‐ tifiable concepts, visualization, Radical Construction Grammar (RCG), Uniform Meaning Representation (UMR), and Interlinear Morphemic Glossing (IMG). The resulting annotation is well suited for multilingual applications, especially for second language acquisition. We have integrated the representation into a web-based language learning environment to allow instructors to design <?page no="136"?> 1 according to the Semantic View of Concepts, concepts are abstract objects, they are the constituents of Fregean propositions 2 https: / / uvi.colorado.edu/ engaging annotation exercises using augmented browsing technology. Based on language exposure, we populate personal idiolectal constructicons that represent the students’ current state of second language comprehension. In our use cases, we focus on the Japanese language because it poses unique challenges for European learners. We also discuss extensions towards transla‐ tion deviation annotations, support for under-resourced languages, and various user interface design choices based on lessons learned in augmentative and alternative communication, colourful semantics, and iconic and international auxiliary languages. 2 Language and Meaning Representation The task of representing concepts 1 (Margolis and Laurence 2007, 2023) through expressions, words, or even unambiguous word senses is part of the research field of lexical semantics (Saint-Dizier and Viegas 2009). Even though there exist several theories of lexical semantics, the field has been dominated by the cognitive semantics paradigm for the last few decades (Geeraerts 2010). The best-known lexical database of semantic relations organized as network is WordNet (Fellbaum 1998). Nouns, verbs, adjectives, and adverbs are grouped into sets of cognitive synonyms, called synsets, each expressing a distinct concept. The WordNet idea has been applied to many languages, resulting in the creation of an Open Multilingual Wordnet (Bond and Paik 2012), which contains wordnets for 28 languages, all linked to the synsets of the original English WordNet. The largest multilingual resource for named entities and more specific terms is certainly Wikipedia. There have been several projects to extract structured content from the information contained in Wikipedia, the two largest being DBpedia (Lehmann et al. 2015) and Wikidata. An ambitious effort to cover WordNet, Wikipedia as well as many other lexical resources is BabelNet (Navigli et al. 2021). A different family of resources started from research on semantic role labeling, e.g. FrameNet (Fillmore et al. 2003), VerbNet (Kipper-Schuller 2005), or PropBank (Pradhan et al. 2022). They provide much more detailed information on verb usage and can be accessed through a Unified Verb Index 2 . A related effort by the BabelNet team is VerbAtlas (di Fabio et al. 2019). Among these resources, PropBank had the greatest impact on the use in meaning representations. The information contained in PropBank is organized in frame files, which are then further disambiguated in rolesets. The connection 136 Werner Winiwarter <?page no="137"?> 3 typology studies languages starting from the diversity of grammatical structures, deriving general patterns found in that diversity 4 https: / / nlp.uniroma1.it/ spring/ from the rolesets to the different surface forms is established by means of alias links. Multimodal enhancements of lexical resources have a long history but only recently gained new momentum due to the strong interest in research on visual question answering or multimodal Large Language Models (LLMs). Regarding the mapping of images to WordNet synsets, there exists the ImageNet collection, which maps approx. 1,000 images to each synset (Deng et al. 2009). A much more influential resource is Wikipedia, which has been increasingly enhanced with visual representations. However, the number of images varies widely across language versions. The most comprehensive recent effort is BabelNet with the annotation tool Babelfy (Moro et al. 2014) and the BabelPic (Calabrese et al. 2020) dataset targeting non-concrete concepts. The use of pictorial illustrations has a long history in language teaching didactics and there exist numerous empirical studies that show their effectiveness at all levels of proficiency (Tahiri 2020). Nonetheless, to the best of our knowledge, we are not aware of any related work with the aim of creating visual representations of meaning representations of sentences. Construction Grammar (Ungerer and Hartmann 2023) is actually a broad family of theories in cognitive linguistics (Croft and Cruse 2004). Their common denominator is the view that constructions are the basic units of language, which are pairings of form and meaning. RCG (Croft 2001, 2022) was designed with typological 3 (Croft 2002) applicability as main motivation. RCG considers word classes and other syntactic structures as language-specific and construc‐ tion-specific. One of the main consequences of Construction Grammar is that the traditional divide between lexicon and grammar is abandoned. The lexicon is replaced by a network of constructions, words being constructions. Most attempts to assemble such a constructicon are in the context of the FrameNet project (Lyngfelt 2018). Whereas the original model was a taxonomical inheri‐ tance network, recent research on usage-based linguistics (Divjak 2019) has led to a multidimensional association network (Diessel 2023). The annotation of sentences with meaning representations has established itself in the last decade as a thriving research field in computational linguistics (Abend and Rappoport 2017). The most influential approach has been the Abstract Meaning Representation (AMR) (Banarescu et al. 2013). There are many parsers available, one popular choice is SPRING 4 . One point of criticism concer‐ ning AMR’s reliance on numbered, not directly interpretable core arguments, A Constructionist Pictorial Approach Towards a Universal Language Representation 137 <?page no="138"?> 5 https: / / umr-tool.cs.brandeis.edu/ 6 an interlinear gloss is a gloss that is placed between lines, e.g. between a line of source text and its translation 7 https: / / www.eva.mpg.de/ lingua/ tools-at-lingboard/ glossing_rules.php is addressed by the WISeR meaning representation (Feng et al. 2023), which maps them to thematic roles. AMR has been recently extended to UMR (Van Gysel et al. 2021). It enhances AMR by adding support for other languages (in particular low-resource languages), and a document-level representation capturing intersentential coreference and temporal/ modal dependencies. UMR fully embraces RCG as a theoretical foundation. First research activities focused on developing the annotation tool UMR Writer 5 . In addition, research work on mapping AMR to UMR has been initiated (Bonn et al. 2023). Whereas the topic of interlinear glossing 6 has been neglected for a long time by natural language processing research, it has a long tradition in linguistics and typology. A recent historical comparative study on glossing practices with a focus on Japan can be found in Cinato et al. (2023). Starting from around 2010 there have been increased activities to collect and work with Interlinear Glossed Text (IGT) (Xia et al. 2015). Well-known results from this effort are the Xigt format (Goodman et al. 2015) and the ODIN database (Xia et al. 2014) with over 200,000 instances of interlinear glosses for more than 1,500 languages. A specific form of interlinear glossing is an IMG (Lehmann 2004). It represents a text in the source language by a string of elements from the target language by ideally mapping each morph of the source language to a morpheme of the target language or grammatical category labels. For a long time, there was no common format resulting in a confusing variety of glossing styles. This has changed because of the Leipzig Glossing Rules 7 . Only recently, there have been some efforts by computational linguists to extend and formalize the guidelines (Mortensen et al. 2023) and to automate interlinear glossing (Samardžić et al. 2015; Zhao et al. 2020). There has even been a first shared task on this topic in 2023 (Ginn et al. 2023). 3 COPAL In this section we present our novel language representation, a COnstructionist Pictorial Approach towards Language (COPAL). It consists of a structural view based on RCG and IMG, and a semantic view based on UMR. As unique concept identifiers, we use PropBank rolesets, WordNet synsets, or Wikipedia page titles. The visual elements of our representation include images for concepts, emoji for roles, and color coding for semantic classes and information 138 Werner Winiwarter <?page no="139"?> 8 the language learning server is implemented in SWI-Prolog (Wielemaker et al. 2012), which is not only predestined for natural language tasks but also provides a scalable Web server solution and libraries for efficiently handling huge RDF and XML files; the server parses sentences by using the linguistic knowledge stored in the constructicons and dynamically generates an HTML page with the annotated sentence, adding new information to the annotation leads to an update of the constructicon, a reparsing of the sentence, and an actualization of the HTML page 9 we will make the source code (including an online demo and documentation) publicly available on Codeberg in the near future; in collaboration with our international partners, we are currently conducting systematic evaluations (McMurry 2016), the first results and insights are being prepared for publication packaging functions. In this chapter we can only present a brief overview of COPAL, for further details about formal aspects, the implementation of personal idiolectal constructicons and the required computational and annotation tools as components of our web-based language learning environment 8 , and the user interaction design in annotation exercises see Winiwarter (2024b) 9 . In RCG, there exist two central comparative concepts: constructions and strategies. A construction (cxn) is any pairing of form and function in any language to express a particular combination of semantic content (sem) and information packaging (inf); a strategy (str) further distinguishes certain characteristics of grammatical form defined in a crosslinguistically consistent way. Definitions of all boldface terms can be found in the glossary of Croft (2022). There are three main information packaging functions (propositional acts): reference, modification, and predication. There are also three main semantic classes: objects, properties, and actions. Figure 1 shows the possible combinations of propositional acts and semantic classes. The prototypical combinations can be seen along the main diagonal, however, any combination of semantic content and information packaging is possible, resulting in a 3x3 matrix of constructions, see Winiwarter (2024a) for more details. Fig. 1: Propositional acts and semantic classes A Constructionist Pictorial Approach Towards a Universal Language Representation 139 <?page no="140"?> 10 https: / / ja.wikisource.org/ wiki/ 透明少女 , https: / / en.wikisource.org/ wiki/ The_Invisibl e_Girl Starting from atomic constructions, constructions can be used as elements to build complex constructions by extending a head with dependents. The two main complex constructions are phrases and clauses. In the following, we illustrate this assembly process in COPAL by using one illustrative example. It is part of a sentence from the Japanese translation 10 of the 1833 tale “The Invisible Girl” by Mary Shelley. The original text reads as “… I was surprised on visiting what seemed a ruined tower, crowning a bleak promontory overhanging the sea, that flows between Wales and Ireland …”. Fig. 2: Example of COPAL representation (Level 1 and 2) Figure 2 shows the first two levels of building the COPAL representation. Thanks to the particular characteristics of Japanese grammar, we can retrace this bottom-up procedure in a strictly left-to-right manner. The simultaneous creation of the semantic view is mirrored, i.e. it is built from right to left. The boxes at the top represent the atomic constructions from the structural view and show their canonical trilinear representations. They consist of three lines: transliterations using Revised Hepburn romanization, IMGs, and idiomatic translations to concepts. For instance, the first word ウェールズ is mapped to the Wikipedia page title Wales, leading to the display of the corresponding visual concept representation below at the right top as a starting point for assembling the semantic view. We retrieve all images from Wikimedia Commons, either through the Wikidata image property or our image database, which currently contains over 3,500 images that have been manually annotated with semantic 140 Werner Winiwarter <?page no="141"?> tags and links to WordNet synsets. All visual elements can be freely customized to account for cross-cultural interpretability. The background colour magenta indicates that the semantic class is object. Fig. 3: Example of COPAL representation (Level 3) The border colour of the construction represents the information packaging function. The colour magenta signals the function reference. In this case, we have a prototypical combination, hence the same colour. The construction label of the resulting construction is shown as a tooltip, it is a referent expression as head of a referring phrase. In the same way, we map アイルランド to the Wikipedia page Ireland. The word と (to) is a member of a short list of common particles. As each of them has many different syntactic functions and no independent semantic content, it is not necessary to add any annotation for such monosyllabic words. The expression の間を consists of three words and is also used like a postposition, therefore we only add the IMG between. A Constructionist Pictorial Approach Towards a Universal Language Representation 141 <?page no="142"?> The second main type of atomic constructions are predicates, i.e., heads of clauses. 流れる is the dictionary form of a Japanese verb and consists of two morphemes: stem 流れ and ending る . This is indicated by a hyphen and NPST (non-past) in the IMG. It is mapped to the ProbBank roleset flow-01, an action painted in cyan. This is again a prototypical pairing; therefore we also use the border colour cyan. The only complex construction at Level 1 is a path event cxn. It is modelled in the semantic view by two PATH-BETWEEN ⌤ relations from flow-01 to the two locations Wales and Ireland. At Level 2, we add 海 , which is mapped to the WordNet synset sea.n.01, and 突き出た , the past tense (PST) of 突き出る . The past tense is modelled in the semantic view as a BEFORE ⏪ relation from project-01 to the special concept DCT (Document Creation Time), coloured in orange, indicating the present moment. The most frequent clause type occurring in this example is the relative clause cxn. At Level 2 it consists of the dependent relative clause … 流れる , which denotes the event that is used to modify the relative clause head 海 in the matrix clause 海に突き出た : projecting towards the sea, that flows between Wales and Ireland. The semantic view contains the role GOAL 🎯 between project-01 and sea.n.01. Fig. 4: Example of COPAL representation (Level 4 and 5) Figure 3 introduces the third main type of atomic constructions: modifiers, which are the heads of attributive phrases. The expression 荒涼とした acts like an adjective and is translated to the property bleak-01, therefore, we have again a prototypical combination indicated by the colour green. From the 142 Werner Winiwarter <?page no="143"?> 11 partial strength, affirmative polarity referent expression 岬 (promontory.n.01) and the dependent modifier 荒涼と した we can assemble the modification cxn bleak promontory. In the semantic view, promontory.n.01 is linked to bleak-01 via the inverse relation THEME-OF 📜 . Inverse relations are indicated by the background colour green. We result in the relative clause cxn rising from a bleak promontory projecting …. In the semantic view, promontory.n.01 is linked to project-01 via a THEME-OF 📜 and to rise-01 by a START 🎬 relation, which indicates the location from which the motion originates. Finally, Figure 4 shows the two Levels 4 and 5. At Level 4, 荒涼 is mapped to ruin-01, an action, which is packaged as a reference, i.e., a ruin as a result of a ruining process. The expression のような is a hedge, i.e., it expresses ambiguity, probability, caution, or indecisiveness (Lakoff 1972). It is an example of an admodifier, a construction that performs the information packaging function of admodification (border colour teal) by qualifying a modifier. At the same time, the grammatical category AT (attributor) in the IMG indicates that the combination with any qualified construction results in an attributive phrase: seemingly ruined or which seems to be ruined. In the semantic view, ruin-01 is linked to the special concept PrtAff 11 via the modal strength relation MODSTR 🏋 . In a second step we form a modification cxn by modifying 塔 with the attributive phrase resulting in: seemingly ruined tower. In the semantic view, the action ruin-01 is linked to the object tower.n.01 by the inverse relation UNDERGOER-OF 🚑 . The final word 訪れて is a coordinative form of the Japanese verb 訪れる , the ending indicates the coordinator (CO) function. It is a deranked predicate, i.e. it can act like a conjunction as coordinator and is colour-coded in brown to indicate this strategy. At Level 5, the deranked relative clause cxn visiting a seemingly ruined tower rising … is used to build an adverbial clause cxn. It consists of a matrix clause 驚いた and an adverbial dependent clause … 訪れて . In this case, it is a temporal relation with a partial or full overlap between child and parent, i.e., I was surprised while visiting …, which is depicted in the semantic view by an OVERLAP relation between surprise-01 and visit-01. To retrace the assembly of our example sentence in this section has been quite a challenge. Fortunately, it is a much more entertaining and intuitive process for students when using our language learning environment. A Constructionist Pictorial Approach Towards a Universal Language Representation 143 <?page no="144"?> 12 très-influent, influential, 有力な 13 en plein Maryland, in the State of Maryland, メリーランド州の真ん中に 4 Extensions and Variations During the last few years we have developed and experimented with numerous extensions to address the specific needs of additional user groups as well as variations on our language representation COPAL regarding different user interface design choices to further improve usability and comprehensibility. In the following, we provide a short overview of translation deviation annotations, support for under-resourced languages, and user interface simplifications based on research on augmentative and alternative communication, colourful seman‐ tics, and iconic and international auxiliary languages. 4.1 Translation Deviation Annotations Professional translators and interpreters but also students of translation, lang‐ uage, and literary studies often face the situation that they have to compare an original document with one or more translations to analyse the differences between them. Therefore, there is an urgent need for tool support to annotate such translation deviations. We have implemented a multilingual annotation solution by using English as anchor language and comparing it with French and Japanese versions (Winiwarter 2024c). As a first use case we use French novels and compare them with their English and Japanese translations. Figure 5 shows some examples of translation deviation annotations (with added tooltip texts). This version is still based on AMR and uses PropBank role glosses for core arguments. The background colour is chosen to indicate the concept identifier type. The image top left contains two insertions. The first one is the word très in the French original, which is missing in the English translation 12 , and is mapped to very.r.01. To keep the view manageable, we only show an emoji, however, the detailed information can still be displayed by clicking on it. The relation for : degree 🌡 is drawn in cyan to indicate the language French. In the same way, we map the additional expressions en plein and 真ん中 to center.n.01 and link it to Maryland by : part 🍰13 . This line is drawn in salmon to indicate both French and Japanese (just Japanese would be orange). The wastebasket 🗑 in the bottom right corner of the concept Perfection indicates that this concept is missing in the respective languages. 144 Werner Winiwarter <?page no="145"?> Fig. 5: Example of translation deviation annotations Another common case is the substitution of one or several concepts in the source language by alternative concepts in the target language. An example of a simple substitution can be seen in the centre below. The concept displayed for the English text pocket-pistols is Pocket_pistol, which is an exact translation of the French original pistolets de poche. However, in Japanese it is translated as 拳銃 , which can be mapped to the concept Pistol. This deviation is indicated by the symbol ⇄ . A more complex situation is depicted on the right in Figure 5. In this case, the original French expression for the taste for military matters is l’instinct militaire. This is indicated by a substitute concept Instinct with the relation : ARG1 (entity changing) from become-01 and a relation : mod to Military. Since the Japanese text offers a precise translation 軍人としての本 能 of the French original, the deviation from the English version concerns both languages. A Constructionist Pictorial Approach Towards a Universal Language Representation 145 <?page no="146"?> 14 https: / / github.com/ erionc/ albnlp 4.2 Under-Resourced Languages We have designed our language representation from the beginning with a broad multilingual applicability in mind. To demonstrate the usefulness of our approach for under-resourced languages, we have recently started some research work on Albanian. Albanian is a very suitable candidate because hardly any resources or tools exist that could be used for NLP applications (see also the AlbNLP 14 repository). This is particularly true regarding bilingual resources and is surprising because Albanian is the official language of Albania and Kosovo and has worldwide over 7 million first-language speakers. One reason that makes the adaptation from other languages so difficult is that the Albanian language is an isolate: it is the only still existing member of the Albanoid branch of the Indo-European families. Albanian is also not covered by DeepL so that we had to fall back to Google Translate. There is also no trained pipeline for Albanian in SpaCy or in any other popular NLP toolkit; not even a syntactic parser or any other more advanced NLP analysis tool is available for Albanian. Fortunately, at least an Albanian morphological analyzer exists as PyPI package uniparser-albanian, which we could use to deal with the complex Albanian morphology to assist in producing the IMG for our language representation. It also contains English word translations to provide hints for word alignment. As transliteration, we use IPA phonemic transcription. Figure 6 shows our language representation for the first half of an example sentence from the Albanian Wikipedia page about Harry Potter. It also show‐ cases a different display style in that we have embedded the semantic view into the structural view by aligning the concepts directly with the constructions. The two formats both have advantages and disadvantages. Therefore, we leave it to the users to choose their preferred display option. 146 Werner Winiwarter <?page no="147"?> 15 https: / / www.openaac.org/ 16 https: / / www.opensymbols.org Fig. 6: Example of Albanian language representation 4.3 Augmentative and Alternative Communication Augmentative and Alternative Communication (AAC) is a research field with a long tradition. Its subject are communication methods to supplement or replace speech or writing for people with impairments in the production or comprehension of these modalities (Hall et al. 2023). Only in the last few years there has been a noticeable increase in research activities on using Natural Language Processing (NLP) for AAC applications, e.g. Gerlach et al. (2024) and Macaire et al. (2024). A recent survey (Andzik and LaMarca 2024) emphasises the importance of multilingual support for AAC clients across all languages spoken at home, at school, or in the community. AAC devices range from low-tech aids like communication books or boards to high-tech systems. One common feature is the use of symbols as visual representation of concepts, which are often presented in a grid format. The OpenAAC 15 initiative works on creating free, open standards for the AAC community, and on sharing free, open-licensed materials and tools. An important part of this effort is OpenSymbols 16 , a huge collection of open-licensed picture symbols for the use in AAC, which currently includes access to more than 50,000 symbols and icons. In our research, we have started to adapt the design principles and experi‐ ence from AAC by replacing our images with OpenSymbols to simplify the representation for novice learners and for the inclusive use by AAC clients. The original image can still be displayed by clicking on the icons. Figure 7 (a) shows the semantic view of the example sentence six people sat round a small table. We also apply a simplified grid representation according to the AAC user interface design, i.e., we place all the concepts and roles on separate quadratic tiles of equal size. A Constructionist Pictorial Approach Towards a Universal Language Representation 147 <?page no="148"?> 17 http: / / www.iconji.com/ Fig. 7: Semantic view using (a) OpenSymbols (b) Colourful Semantics 4.4 Colourful Semantics Colourful Semantics (Bryan 1997) is an AAC approach, which specifically targets clients with difficulties in sentence planning and word order. It cuts up a sentence into its parts according to their thematic roles and colour codes them. The resulting pieces are simply named after the question words that are used to ask for them. As can be seen in Figure 7 (b), this visualization paradigm can be nicely integrated into our language representation. For the ease of the reader we added the question words and concept identifiers. 4.5 Iconic and International Auxiliary Languages A final research field with a very long history from which we could draw many insights and inspirations is iconic and international auxiliary languages. The modern interpretation of the term iconic languages (Francese et al. 2017) is that they are visual languages making an extensive use of images and icons. In particular, the emergence of mobile devices generated a need for new simple modalities. One such popular pictographic communication system was iConji 17 , however, further development stopped in 2012. A recent related effort is IKON 148 Werner Winiwarter <?page no="149"?> 18 https: / / iconlang.net/ 19 you come to me in order to talk or more literally: talk / do (action) / for, in order to (purpose) / you / I, me / to (destination) / go, come / do (action) 20 property 🤲 , action 🎬 21 literal meaning 👌 , abstract interpretation ⭐ , category 〰 , opposite meaning ⚡ 22 https: / / www.globasa.net/ eng 23 https: / / en.kokanu.com/ 24 literally: using / fire / direct object / to consume (food) / verb / to ready (a prepared ent‐ ity) / subject / I (Meloni et al. 2022), which introduces several strategies to compose compound icons. In his paper, Kramer (2023) extended this line of work towards a graphical representation of complete sentences by presenting a blueprint for a new iconic language Icono. Most of these ideas were incorporated in one of the latest efforts: Iconic 18 , which only uses emoji for its visual presentation. Its vocabulary contains more than 3,000 English words. A simple example sentence in Iconic is: 💬🎬🎁▫👉▫👈📬▫🚶🎬▪ 19 . An interesting feature of Iconic are suffixes that change the information packaging to a property or an action 20 . In the same way, there are suffixes for literal meaning, abstract interpretation, category, or opposite meaning 21 . International Auxiliary Languages (IAL) aim to enable the communication between people from different nations, who do not share a common first lang‐ uage. For a long time, the success of Esperanto overshadowed any activities on novel IALs. Recent efforts particularly address the dominance of Indo-European languages, e.g. Globasa 22 , which takes its vocabulary from the about 17 most spoken languages by using a specific selection process. Another interesting IAL is Toki Pona (Lang 2014). Its underlying feature is minimalism so that the original proposal contained a vocabulary of only 120 words. A recent extension of Toki Pona towards the use as IAL is Kokanu 23 with a dictionary of currently 372 entries and consistent rules to change the information packaging of any noun, verb, or modifier. In our research, we have done some first experiments with an iconic version of Kokanu. Figure 8 shows the simple example sentence I prepare food using fire  24 . Fig. 8: Example of iconic Kokanu A Constructionist Pictorial Approach Towards a Universal Language Representation 149 <?page no="150"?> 5 Conclusion In this chapter, we have taken the reader on a journey towards a universal language representation. We have been able to develop our COPAL approach by collecting many seeds from previous research activities. Starting with support for Japanese-language learners, we have gradually extended and adapted our representation to meet the specific needs of other user groups. An important lesson we have learnt in our quest is that there is no single perfect representation but only optimal personalisation. User constructicons are promising candidates for modelling mental lexicons, especially if they incorporate associations for the visual modality. Thus, not only does cognitive linguistics benefit from the latest technological advances in LLMs, but cognitive linguistics is also a key component for the next generation of AI. Bibliography Abend, Omri/ Rappoport, Ari (2017). The state of the art in semantic representation. In: Barzilay, Regina/ Kan, Min-Yen (Eds.). Proc. of the 55th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). Vancouver: Association for Computational Linguistics, 77-89. Andzik, Natalie R./ LaMarca, Hannah N. (2024). Multilingualism and augmentative and alternative communication: A review of the literature. Perspectives of the ASHA Special Interest Groups, 9(3), 642-651. Banarescu, Laura/ Bonial, Claire/ Cai, Shu/ Georgescu, Madalina/ Griffitt, Kira/ Hermjakob, Ulf/ Knight, Kevin/ Koehn, Philipp/ Palmer, Martha/ Schneider, Nathan (2013). Abstract Meaning Representation for sembanking. In: Nancy Ide/ Adam Meyers (Eds.). Procee‐ dings of the 7th Linguistic Annotation Workshop and Interoperability with Discourse (LAW VII & ID). Sofia: Association for Computational Linguistics, 178-186. Bond, Francis/ Paik, Kyonghee (2012). A survey of wordnets and their licenses. In: Christiane Fellbaum/ Piek Vossen (Eds.). Proceedings of the 6th Global WordNet Conference (GWC 2012). Matsue: Global WordNet Association, 64-71. Bonn, Julia/ Oepen, Stephan/ Abend, Omri/ Rappoport, Ari (2023). Mapping AMR to UMR: Resources for adapting existing corpora for cross-lingual compatibility. In: Tomaž Erjavec/ Darja Fišer (Eds.). Proceedings of the 21st International Workshop on Treebanks and Linguistic Theories (TLT 2023). Prague, 74-95. Bryan, Alison (1997). Colourful semantics. In: Chris Code/ David J. Müller (Eds.). Lang‐ uage Disorders in Children and Adults: Psycholinguistic Approaches to Therapy. London: Whurr, 143-161. Calabrese, Agostina/ Agirre, Eneko/ Baroni, Marco/ Biemann, Chris/ Chersoni, Emma‐ nuele/ Fernandez, Raquel/ Lenci, Alessandro/ Moro, Andrea/ Vecchi, Eva Maria Marelli 150 Werner Winiwarter <?page no="151"?> (2020). Fatality killed the cat or: BabelPic, a multimodal dataset for non-concrete con‐ cepts. In: Dan Jurafsky/ Joyce Chai/ Natalie Schluter/ Joel Tetreault (Eds.). Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics (ACL 2020). Online: Association for Computational Linguistics, 4680-4686. Cinato, Franck/ Lahaussois, Aimée/ Whitman, John B. (2023). Glossing Practice: Compa‐ rative perspectives. Lanham: Rowman & Littlefield. Croft, William (2001). Radical Construction Grammar: Syntactic theory in typological perspective. Oxford: Oxford University Press. Croft, William (2002). Typology and Universals. Cambridge: Cambridge University Press. Croft, William (2022). Morphosyntax: Constructions of the world’s languages. Cam‐ bridge: Cambridge University Press. Croft, William/ Cruse, David Alan (2004). Cognitive Linguistics. Cambridge: Cambridge University Press. Deng, Jia/ Dong, Wei/ Socher, Richard/ Li, Li-Jia/ Li, Kai/ Fei-Fei, Li (2009). ImageNet: A large-scale hierarchical image database. In: Carl Vondrick/ Deva Ramanan/ Pietro Perona (Eds.). IEEE Computer Society Conference on Computer Vision and Pattern Recognition (CVPR 2009). Miami: IEEE, 627-637. Diessel, Holger (2023). The Constructicon: Taxonomies and Networks. Cambridge: Cambridge University Press. Di Fabio, Andrea/ Conia, Simone/ Montemagni, Simonetta (2019). VerbAtlas: A novel large-scale verbal semantic resource and its application to semantic role labeling. In: Kentaro Inui/ Jing Jiang/ Vincent Ng/ Xiaojun Wan (Eds.). Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natural Language Processing (EMNLP-IJCNLP 2019). Hong Kong: Association for Computational Linguistics, 627-637. Divjak, Dagmar (2019). Frequency in Language: Memory, Attention and Learning. Cambridge: Cambridge University Press. Fellbaum, Christiane (1998). WordNet: An Electronic Lexical Database. Cambridge, MA: MIT Press. Feng, Lydia/ Li, Tianyu/ Yu, Mo/ Wang, Danqi (2023). Widely Interpretable Semantic Representation: Frameless Meaning Representation for Broader Applicability. CoRR abs / 2309.06460. Fillmore, Charles J./ Johnson, Christopher R./ Petruck, Miriam R. L. (2003). Background to FrameNet. International Journal of Lexicography, 16(3), 235-250. Francese, Rita/ Pace, Giuseppina/ Passero, Ignazio (2017). Iconic languages: Towards end-user programming of mobile applications. Journal of Visual Languages and Computing, 38. 1-8. Geeraerts, Dirk (Ed.). (2010). Theories of Lexical Semantics. Oxford: Oxford University Press. A Constructionist Pictorial Approach Towards a Universal Language Representation 151 <?page no="152"?> Gerlach, Johanna/ Schulz, Stefan/ Mader, Luise (2024). A concept-based approach for translation of medical dialogues into pictographs. In: Nicoletta Calzolari/ Philippe Bla‐ che/ Khalid Choukri/ Christopher Cieri/ Thierry Declerck/ Sara Goggi/ Koiti Hasida/ Hi‐ toshi Isahara/ Bente Maegaard/ Joseph Mariani/ Asunción Moreno/ Jan Odijk/ Stelios Piperidis (Eds.). Proceedings of the 2024 Joint International Conference on Computa‐ tional Linguistics, Language Resources and Evaluation (LREC-COLING 2024). Torino: ELRA and ICCL, 233-244. Ginn, Michael/ Cotterell, Ryan/ Sylak-Glassman, John/ Hulden, Mans (2023). Findings of the SIGMORPHON 2023 shared task on interlinear glossing. In: Ryan Cotterell/ Timo‐ thy Baldwin/ Clara Vania (Eds.). Proceedings of the 20th SIGMORPHON Workshop on Computational Research in Phonetics, Phonology, and Morphology. Toronto: Association for Computational Linguistics, 186-201. Goodman, Michael Wayne/ Xia, Fei/ Schneider, Nathan (2015). Xigt: Extensible interlinear glossed text for natural language processing. Language Resources and Evaluation, 49(2), 455-485. Hall, Nerissa/ Biedermann, Britta/ Erickson, Krista/ Light, Janice (2023). Fundamentals of AAC: A Case-Based Approach to Enhancing Communication. San Diego, CA: Plural Publishing. Kipper-Schuler, Karin (2005). VerbNet: A Broad-Coverage, Comprehensive Verb Lexicon. Philadelphia: University of Pennsylvania. Kramer, Peter (2023). Icono: A universal language that shows what it says. Frontiers in Psychology, 14. Lakoff, George (1972). Hedges: A study in meaning criteria and the logic of fuzzy concepts. Journal of Philosophical Logic, 2(4), 458-508. Lang, Sonja (2014). Toki Pona: The Language of Good. Tawhid. Lehmann, Christian (2004). Interlinear morphemic glossing. In: Geert Booij/ Christian Lehmann/ Joachim Mugdan/ Stavros Skopeteas (Eds.). Morphologie: Ein internationa‐ les Handbuch zur Flexion und Wortbildung. Berlin/ New York: De Gruyter Mouton, 1834-1857. Lehmann, Jens/ Isele, Robert/ Jakob, Max/ Jentzsch, Anja/ Kontokostas, Dimitris N./ Men‐ des, Pablo N./ Hellmann, Sebastian/ Morsey, Mohamed/ van Kleef, Patrick/ Auer, Sö‐ ren/ Bizer, Christian (2015). DBpedia - A large-scale, multilingual knowledge base extracted from Wikipedia. Semantic Web, 6(2), 167-195. Lyngfelt, Benjamin (2018). Constructicography: Constructicon Development Across Languages. Amsterdam/ Philadelphia: John Benjamins. Macaire, Cécile/ Estève, Yannick/ Lecouteux, Benjamin/ Besacier, Laurent (2024). A multimodal French corpus of aligned speech, text, and pictogram sequences for speech-to-pictogram machine translation. In: Nicoletta Calzolari/ Philippe Bla‐ che/ Khalid Choukri/ Christopher Cieri/ Thierry Declerck/ Sara Goggi/ Koiti Hasida/ Hi‐ 152 Werner Winiwarter <?page no="153"?> toshi Isahara/ Bente Maegaard/ Joseph Mariani/ Asunción Moreno/ Jan Odijk/ Stelios Piperidis (Eds.). Proceedings of the 2024 Joint International Conference on Computa‐ tional Linguistics, Language Resources and Evaluation (LREC-COLING 2024). Torino: ELRA and ICCL, 839-849. Margolis, Eric/ Laurence, Stephen (2007). The ontology of concepts - abstract objects or mental representations? Noûs, 41(4), 561-593. Margolis, Eric/ Laurence, Stephen (2023). Concepts. In: Edward N. Zalta (Ed.). The Stanford Encyclopedia of Philosophy. Stanford: Stanford University. McMurry, Benjamin/ Smith, Jennifer/ Lee, Daniel (2016). An evaluation framework for CALL. TESL-EJ, 20(2), 1-31. Meloni, Laura/ Rossi, Fabio/ Conti, Alessandro (2022). Beyond emojis: An insight into the IKON language. In: Yuval Pinter/ Louise-Amélie Schaeffer (Eds.). Proceedings of the Fifth International Workshop on Emoji Understanding and Applications in Social Media. Barcelona: Association for Computational Linguistics, 11-20. Moro, Andrea/ Colavizza, Giovanni/ van Erp, Marieke (2014). Entity linking meets word sense disambiguation: A unified approach. Transactions of the Association for Computational Linguistics, 2. 231-244. Mortensen, David R./ Cotterell, Ryan/ Hulden, Mans (2023). Generalized glossing guide‐ lines: An explicit, humanand machine-readable, item-and-process convention for morphological annotation. In: Ryan Cotterell/ Timothy Baldwin/ Clara Vania (Eds.). Proceedings of the 20th SIGMORPHON Workshop on Computational Research in Phonetics, Phonology, and Morphology. Toronto: Association for Computational Linguistics, 58-67. Navigli, Roberto/ Ponzetto, Simone Paolo/ Del Corro, Luca (2021). Ten years of BabelNet: A survey. In: Subbarao Kambhampati/ Patrick Gallinari (Eds.). Proceedings of IJCAI 2021. Montreal: IJCAI. 4559-4567. Pradhan, Sameer/ Hajic, Jan/ Poesio, Massimo (2022). PropBank comes of age - Larger, smarter, and more diverse. In: Steven Bird/ Chaitanya Shivade (Eds.). Proceedings of *SEM 2022. Dublin: Association for Computational Linguistics, 278-288. Saint-Dizier, Patrick/ Evelyne Viegas (Eds.) (2009). Computational Lexical Semantics. Cambridge: Cambridge University Press. Samardžić, Tanja/ Grondahl, Bjørn/ van den Bosch, Antal (2015). Automatic interlinear glossing as two-level sequence classification. In: Dan Flickinger/ Nancy Ide (Eds.). Proceedings of LaTeCH 2015. London: Association of Computational Linguistics, 68-72. Tahiri, Shejla (2020). The impact of pictures on second language acquisition. SEEU Review, 15(2), 126-135. Unger, Tobias/ Stefan Hartmann (2023). Constructionist Approaches: Past, present, fu‐ ture. Cambridge: Cambridge University Press. A Constructionist Pictorial Approach Towards a Universal Language Representation 153 <?page no="154"?> Van Gysel, Jens E. L./ Mihalcea, Rada/ Weikum, Gerhard (2021). Designing a Uniform Meaning Representation for natural language processing. KI, 35(1), 343-360. Wielemaker, Jan/ Wolter, Frank/ Schrijvers, Tom (2012). SWI-Prolog. Theory and Practice of Logic Programming, 12(1), 67-96. Winiwarter, Werner (2024a). CLAVELL - Cognitive Linguistic Annotation and Visuali‐ zation Environment for Language Learning. In: Michael Johnston/ Christine Howes (Eds.). Proceedings of CogALex VIII. Vienna: European Association for Lexicography, 1-13. Winiwarter, Werner (2024b). RUVA - a Radical Universal Visual Annotation for web-ba‐ sed language learning. In: Markus Strohmaier/ Elena Simperl (Eds.). Proceedings of iiWAS 2024. Brussels: Association for Computing Machinery, 1-16. Winiwarter, Werner (2024c). VOLARE - Visual Ontological Language Representation. In: Michael Johnston/ Christine Howes (Eds.). Proceedings of DMR 2024. Vienna: European Association for Lexicography, 54-65. Xia, Fei/ Wang, Jian/ Liu, Yang (2014). Enriching ODIN. In: Nicoletta Calzolari/ Katrin Erk (Eds.). Proceedings of LREC 2014. Reykjavik: ELRA, 3151-3157. Xia, Fei/ Wang, Jian/ Liu, Yang (2015). Enriching, editing, and representing Interlinear Glossed Text. In: Kam-Fai Wong/ Laurence Simoncini (Eds.). Proceedings of CICLing 2015. Bethlehem: Springer, 32-46. Zhao, Xingyuan/ Chen, Wei/ Li, Ming (2020). Automatic interlinear glossing for under-re‐ sourced languages leveraging translations. In: Alexander Gelbukh/ Michael Strube (Eds.). Proceedings of COLING 2020. Barcelona: Association of Computational Lingu‐ istics, 5397-5408 154 Werner Winiwarter <?page no="155"?> So much technology, so little time Dragoș Ciobanu Abstract: This chapter examines the translation industry’s rapid techno‐ logisation which has been necessary in order to cope with soaring content types and volumes, as well as increasingly complex content authoring and distribution pipelines. Beyond efficiency, the chapter foregrounds human impact, urging people-first design, ergonomic workflows, and transparency about MT/ LLM use and human involvement. It challenges simplistic productivity and edit-distance metrics, showing wide inter-in‐ dividual variability and the need to value cognitive effort and quality outcomes. Using examples from the LT-LiDER and HAITrans Imminent projects, the chapter highlights how technologies such as automatic speech synthesis (text-to-speech) and automatic speech recognition (dic‐ tation/ speech-to-text), combined with a re-centering of human expertise, enable ergonomic, effective, and ethical integrations of AI in language services workflows. 1 Introduction Whether the language services industry will continue to grow by at least 5 % each year until 2029, as Nimdzi Insights suggests (Varga and Groot Nibbelink 2025), or will be on an inflationadjusted downward trend according to the Common Sense Advisory (CSA) Research (Lommel 2025) and the European Language Industry Survey (ELIS 2025), the reality is that people, products, and services still need to reach and engage with a multilingual, multicultural audience. As our increasingly globalised, digitised, and better-connected world keeps producing a staggering volume of multimedia content every day to suit a variety of purposes, this poses significant challenges to both content creators and the language services industry which is constantly adapting to support them. <?page no="156"?> 1 https: / / eur-lex.europa.eu/ eli/ reg/ 2024/ 1689/ oj/ eng At first view, this tsunami of content should represent excellent news for the language specialists delivering services such as translation, interpreting, revision, subtitling, localisation, transcreation, editing, or terminology extraction and management. However, it is increasingly difficult to find evidence of enthusiasm towards the future in their midst. On the contrary, “for the first time in ELIS history, participating language companies expect not only their own activity to decline further, but also the global language industry and in particular their local language service market” (ELIS 2025: 5). Although areas like audiovisual translation and interpreting are still reported to thrive, they are part of a small group of exceptions rather than the rule (ibid.). Other relatively safe tasks include “Machine translation (MT) and post editing, data services, prompt engineering and pure technology offerings” (Varga and Groot Nibbelink 2025). These, however, often either dispense with qualified language specialists’ input altogether or, in combination with language service providers (LSPs) automating their workflows and adjusting their service offerings to keep up with market changes, require language specialists to embrace technological change and adapt their skill sets significantly in order to complete tasks which are increasingly different from the ones most of them trained for. What is more, the “quick evolution of artificial intelligence (AI) tools” (ibid.), which has been fuelling these areas, has also led to even bigger time and financial pressures on most industry members. With some exceptions, of course, new technologies have not delivered all the benefits which language specialists have been told to expect. In fact, their levels of job satisfaction have been decreasing for several years, alongside their perceptions of how desirable their profession is (Sakamoto et al. 2024; Moorkens 2020a; 2020b; Vieira 2020). One of the results of combining content growth with a more dynamic inter‐ pretation of what quality represents and with increased confidence in language technologies such as neural machine translation (NMT), large language models (LLMs), and quality estimation (QE) which are used individually or combined in translation and translation-related tasks is that “A shocking amount of the web is machine translated” (Thompson et al. 2024). As the 2024 EU Artificial Intelligence Act 1 starts making an impact and content providers become more transparent about their creation methods, it may also surprise us to finally see how much content is not only translated automatically, but also created in the same manner. 156 Dragoș Ciobanu <?page no="157"?> 2 https: / / lt-lider.eu/ Will this new information change creation and consumption trends to reflect the numerous ethical debates on the topic, though? Will the human aspect recover its former importance? Time will tell and opinions are split: on the one hand, since the advent of NMT in 2016, “Survey data, commentary on the eco‐ nomic prosperity of the [language services] industry, and claims of technological breakthroughs are masking the lived experience of professionals” (Lambert and Walker 2024: 100). Several industry surveys (ELIS 2025; Inbox Translation and Institute of Translation and Interpreting (ITI) 2023; SFT 2022) have been warning for several years about the profession becoming less sustainable: “The downward pricing trend that started in 2023 became more visible in 2024. Independent professionals expect further price erosion in 2025.” (ELIS 2025: 13). Nevertheless, discussions about fairer working conditions quickly become purely academic as LSPs hasten to point out that, at least for some of the tasks they intermediate, their clients are not willing to pay more and would surely choose another provider for the same or even lower rates. On the other hand, the good news is that, at least in some corners of the language services industry, attitudes are starting to change. Between April and June 2024, the Language and Translation: Literacy in digital environments and resources (LT-LiDER) 2 project team interviewed 29 experts about industry realities, trends, needs, and future. Given that, as Interviewee 5 (i5) put it, “[Ar‐ tificial Intelligence] AI is definitely driving the agenda. Everyone wants to incorporate it and feature the latest advancements”, there was consensus among the interviewees that the “dynamic technological landscape necessitates a high degree of flexibility and a commitment to continuous learning among translators” (Secară et al. 2025). Interestingly, this was not presented as yet another coercive imposition on language specialists, but rather as a strategic move which would position them well for those companies which, disappointed by the reduced deliverables of the AI-first bandwagon, were returning to a people-first mindset: We have also seen it with some companies that at first said, hey, let’s stop having editors, and now they are coming back; when they are seeing the problems they are having, they are coming back to hiring humans. (LT-LiDER: i15) At the end of the day, in addition to the familiar recipe for thriving in the language services industry by specialising and by acquiring direct clients in the premium market, remaining curious and proactive about, as well as familiar with So much technology, so little time 157 <?page no="158"?> the benefits and drawbacks of new technologies, will help language specialists weather the storms ahead. 2 Taming complexity through expert-driven automation Undoubtedly, in a few short decades, the language services industry has become fascinatingly diverse and complex. With a bit of curiosity about what happens before, after, and in parallel with the tasks in which language specialists are involved, one can start to understand why, in some scenarios, technology is such a vital component of the multifaceted efforts to scale up operations. Automation, for example, has been simplified and has become equivalent in some circles with low rates and unreasonable demands addressed to language specialists, but that is a consequence of hype in the attention economy and misimplementations, rather than its raison d’être. In people-first rather than AI-first enterprises, technology is deployed to empower and support (know‐ ledge) workers rather than to replace them, and automation remains a necessity as content volumes are growing and deadlines are reducing. In his ELEVATE 2025 presentation, Scott Cahoon (2025) illustrated how the in-house localisation team at DELL, even after some initial automation and efficiencies brought about by their previous translation management system (TMS), still had to somehow manage 216 customers, 1,381 localisation project templates, 254 workflows, four vendors for language services, 65 pricing models, and also had to maintain 20 integrations with separate content sources. At the time of the presentation, as a result of sustained focus on automation and structural simplifications, their number of internal customers had dropped to 2, the project templates to 43, the workflows to 22, the vendors to 2, the pricing models to 5, and the integrations to 12. With the exception of the number of integrations, which was projected to increase as more content sources are plugged into their latest TMS (currently Trados), their team still had more work to do to reduce the number of project templates to their target of 30, and the number of workflows to 10. The fact that such complexities exist or develop as companies expand, but go largely unacknowledged even by members of the language services industry, let alone the wider public, remains a source of frustration and should be addressed more systematically. Even back in 2018, not just translators were dissatisfied with how their profession was perceived and remunerated, but localisation project managers, too. Interestingly, they had similar suggestions to those frequently made by language specialists for ways to prevent or at least delay professionals from leaving the industry: “Make people more aware of how much 158 Dragoș Ciobanu <?page no="159"?> work is behind a simple translated document, have respect for others’ job/ work” (Bond 2018). Because of the reality of more or less well-informed clients with more or less reasonable expectations, a whole lot of man-made chaos still characterises a substantial number of localisation projects: content with plenty of ambiguities in surprising file formats, technologies which do not in fact talk to each other as they should, unannounced content updates, or sudden changes in project scope and, essentially. As a result, resources are constantly poured into technologies which should alleviate some of these pain points. The aim is always to reduce the potential for human error in complex workflows and streamline processes by creating or repurposing one or more technologies, but the compromises which are inevitably present in these solutions can affect language specialists and end-users, too, sometimes disproportionally. Fortunately, some of these misimplementations can also be remedied at least to some extent by listening to the specialists and the end-users again: Why do I believe linguists are indispensable? It’s because these developments aren’t always made with an understanding of localization or translation in mind. For ins‐ tance, there’s a tendency to say, “Let’s use generative AI for automatic post-editing”, without fully grasping the intricacies of a style guide, or its components. (LT-LiDER: i5) Historically, language specialists have been fairly open to (and some even proac‐ tive about) integrating new technologies into their workspaces: from translation memories (TM) to electronic dictionaries and multilingual termbases, quality assurance (QA) checks, and even machine translation (MT) in some domains and language pairs, especially when the NMT models in question had been expertly fine-tuned and were appropriate for that particular content. When additional tools matured - such as desktop-based computer-assisted translation (CAT) tools evolving into cloud-based TMS, or automatic speech recognition (ASR) and even automatic speech synthesis becoming usable for several languages -, they were adopted to optimise services without affecting product quality (Ciobanu 2016; Ciobanu et al. 2019; Ciobanu et al. 2024). Dictation with ASR tools in particular was mentioned by LT-LiDER interviewees for both improving the ergonomics of the language specialist’s work‐ place and as an effective approach to maintain end-product quality while also coping with shorter deadlines: “It also increases my productivity a lot, and dictating the translation makes me feel that it is already better than writing it, as it sounds more natural.” (LT-LiDER: i15) So much technology, so little time 159 <?page no="160"?> However, we should keep in mind at all times that “Technology […] can have a positive and/ or negative impact on translators’ sense of esteem, belonging, financial well-being, and wider professional-existential concerns” (Lambert and Walker 2024: 99). When translators are trusted to choose the tools and workflows best suited to their clients’ interests, durable professional relationships develop. On the other hand, when their clients change workflows by integrating more translation technologies without always consulting those ultimately using them, a growing sense of disempowerment unsurprisingly settles in among language specialists. What may look exciting in a research lab or top the state-of-the-art scores in a computer science competition could be perceived as patronising and offensive by real-world users because of its obvious unfinished nature. Speaking to the LT-LiDER project, an NMT researcher synthesised the mindset which some developers urgently need to change: “When we, as researchers, go to deploy systems, 9 times out of 10 we don’t think about the target community. We think, they don’t have anything right now, so anything is better than nothing.” (LT-LiDER: i2) At the end of the day, a tool which solves a narrow translation problem, or may even be acceptable for a lay user who has no other options, is not always automatically acceptable or even helpful to a language specialist. Given the vast diversity and complexity of communicative contexts, most tools simply do not scale well, and expecting them to do so is rather too optimistic at this time. A much more sensible approach is to build trusted partnerships with proactive language specialists who have the experience and professionalism needed to track the real achievements of current technologies relevant for their clients’ needs. 3 Tracking productivity Ever since MT came on the scene, the question has always been whether new workflows incorporating it would make translators more efficient. Even though the ALPAC report (1966) put a few spokes in MT development efforts by stating that MT quality did not justify the investment, and dictating a translation was faster at that time than correcting the output of machine translation systems, MT researchers did not down tools. As the technology progressed, regular experiments were conducted, pitching translators not using MT against translators using it. Slowly, especially after the MT paradigm changed from rule-based machine translation (RBMT) to statistical machine translation (SMT), certain objective benefits started being recorded although, subjectively, translators did not acknowledge them: 160 Dragoș Ciobanu <?page no="161"?> Although for all the language pairs users perceived MT post-editing to be slower, in fact it proved to be a faster option than manual translation for two translation directions out of four, i.e. for Dutch to English, and (marginally) for English to German. (Gaspari et al. 2014) Moreover, since the majority of translators were by that time already using translation memories within their CAT tools, productivity assessments were updated to reflect that: translators have higher productivity and quality when using machine-translated output than when translating without it, and […] this productivity and quality is not significantly different from the values obtained when processing fuzzy matches from translation memories in the range 85-94 percent. (Guerberof Arenas 2014) Interestingly, some of these findings appear to hold true for the latest MT paradigm - NMT -, as well: linguists using neural adaptive MT engines spend a similar amount of time editing MT suggestions as they do editing higher fuzzy match ranges: on average, it takes 1.59 seconds per word to edit MT suggestions, while it takes 1.64 seconds per word to edit 85%-94% fuzzy matches. (Matecat n.d.) This data can be converted into an expected productivity of over 2,200 words/ hour which represents, especially in the case of some International Organisati‐ ons, more than what their in-house translators are expected to work on in a full day. Even though exact figures vary according to language pairs, domains, and experiment settings, “an improvement of around 20 % in efficiency by using PE is usually regarded as acceptable” (Yamada 2023: 48). In addition, due to the multitude of decontextualised and simplified success stories promoting the benefits of NMT (and, lately, also LLMs), even language departments within companies, or national and international organisations are being impacted, although historically they used to be fairly sheltered from the evolution of the market. Specifically, “Time pressure and the position of the department within the organisation are two new but very important challenges” (ELIS 2025: 24). The reality that the output quality of NMT models and LLMs, alongside the reliability of QE algorithms, all differ between languages and domains (usually going down from an artificially-evaluated score the more specialised the domain and the more resource-poor the language pair) make it less often into industry headlines. Such precise-looking productivity figures are made possible by tracking tools integrated more or less overtly into popular CAT tools and TMS, aiming to quantify the edit time and effort spent by the language specialists. When So much technology, so little time 161 <?page no="162"?> the difference in terms of characters or words deleted, introduced, moved, or replaced is calculated between the final target segment and its starting version (i.e. nothing in the case of the increasingly rare translations from scratch, or something in the case of revising another translator’s work, or checking and correcting TM matches, TM fragments assembled automatically, or MT/ LLM hypotheses), that is reported as the edit distance. While these metrics were initially presented as assisting project planning and management, it did not take long before they started being used for (under)remunerating translators. However, calculating a linguist’s time and effort only based on this information represents a major simplification of what translation actually entails. First of all, the edit distance score disregards the possibility of several versions of each target segment which the translator may iterate through before settling on the final translation as he/ she goes through several drafts which he/ she self-revises either once or more times, depending on working conditions. If these do not allow time for self-revision, the client should be aware of that and should be willing to take the risk of using a translation which not even the translator looked at a second time. Even when no changes are made to a target segment suggestion, both source and target segments still need to be read closely and considered alongside one or more of the following: the surrounding context; termbase suggestions; alternative TM matches or MT/ LLM equivalents; possible automatic quality assurance (QA) flags and messages; metadata related to the target segment (such as its source, creation and modification dates, client, or domain associated with it); relevant external resources and background information displayed within or outside the CAT tool; and now increasingly detailed LLM-generated explanati‐ ons for (parts of) the source text, as well as automatic evaluations of the target segment suggestion. Needless to say, considering all these data-points needs time and a significant degree of attention to detail and familiarity with all the technologies embedded in or adjacent to the CAT tool/ TMS/ editing environment used for the task. Despite the age-old promise that new technologies will free up translators’ time for creative tasks or leisure, the reality is that the “new systems that provide rich interactivity force translators to swiftly navigate a wealth of sources of information, requiring agility and increased confidence from them” (do Carmo and Moorkens 2022). Secondly, regarding edit time, although not enough translation studies expe‐ riments are explicit about this, translators do (and need to) spend time away from the source and target segments. They perform concordances in their CAT tools or on external corpus-management platforms, look up terminology, or 162 Dragoș Ciobanu <?page no="163"?> 3 https: / / haitrans.univie.ac.at/ research/ imminent/ 4 See which segments are excluded when calculating Post-Editing Effort at https: / / guid es.matecat.com/ quality-report-in-matecat check a variety of preferred external resources, among many others, as already mentioned above. In one of our experiments 3 (Ciobanu et al. 2024), translators (n=21) per‐ forming two PEMT tasks in silence (the silent condition) spent on average 19% of each task time (median=18%, SD=5%) conducting an average of 2.76 task-relevant searches in external resources (median=1, SD=3.42). They were post-editing English source texts of between 342 and 374 words which had been machine-translated into German. When they were able to listen to the source and target segments being “spoken out” by an artificial voice in the experiment CAT tool (Matecat) in two additional PEMT tasks (the sound condition), they spent on average 18 % of each task time (median=18%, SD=5%) conducting an average of 3.57 task-relevant searches in external resources (median=2, SD=3.85). When this necessary additional time is taken into account, together with the longer pauses which translators invariably make when working out how to improve the MT output (and which Matecat, as well as other CAT tools, discard outside the threshold 4 of under 0.5 and over 25 seconds/ word), the final productivity figures differ significantly. Those displayed by Matecat at the end of each task in our PEMT project are similar to the ones reported by the parent company translated above - 1.8 seconds per word on average in the silent condition (median=1.43s, SD=1.4s), and 1.74 seconds per word on average in the sound condition (median=1.33s, SD=1.34s). Nevertheless, they are 2.8 and 2.6 times lower respectively than the ones recorded in our eye-tracking experiment, which also includes the time spent by post-editors consulting external resources: 4.03 seconds per word on average in the silent condition (median= 4.26s, SD=1.21s), and 4.58 seconds per word on average in the sound condition (median= 4.81s, SD= 1.35s). These productivity figures translate into an average of 981.07 words/ hour (median= 844.10, SD= 313.42) in the silent condition, and 874.88 words/ hour (median= 784.82, SD= 310.33) in the sound condition, which are comparable to the averages reported in Terribile (2023: 13). However, average values do not tell the full story. As the standard deviation (SD) values suggest, there is important variation between participants which also needs to be considered when evaluating the impact of technologies on language specialists. In Terribile’s study, for the EN-DE language pair, PEMT productivity ranged between 480 and 1,310 words per hour. In our experiment, the ranges were So much technology, so little time 163 <?page no="164"?> 540-1,616.71 in the silent condition and 530.5-1,872.68 in the sound one. Moreo‐ ver, some translators were faster in the sound condition, but others were not and, overall, productivity dropped when post-editors could listen to the source and target segments (Ciobanu et al. 2024). In other words, it is a reality that some translators derive more productivity benefits than others from new technologies or new ways of working. It is therefore necessary to take with a pinch of salt some of the industry’s reports about approaching the milestone of processing 1 word/ second, and conduct valid and replicable task-specific experiments instead. Moreover, it would be useful for all research looking at productivity to measure additional aspects, such as the quality of the final output, as well as the participants’ potential change in attitudes. In our experiment, the average quality of the PEMT work done in the sound condition was higher than in the silent condition, but the difference was not statistically significant (Ciobanu et al. 2024). Future work will report on a more granular impact assessment which takes into account individual MQM categories (Lommel et al. 2014) rather than just the final scores. Regarding participants’ preferences, “14 participants said they preferred to post-edit [in the sound condition]. Moreover, an even higher proportion (19 out of the 21) said that they would use [the sound condition] in PEMT again” (Ciobanu et al. 2024). 4 An optimistic technology (and speech-)enabled future? Although there is a fair amount of agreement about the growing importance of technology which, as already mentioned in the Introduction, has not brought benefits to all language specialists, some of the LT-LiDER interviewees shared a more optimistic outlook: Having gone through a phase where everything was needed very fast and at low fees with little time for specialisation, we are returning to a specialised space where clients are willing to pay for good content, specialised content aimed at specific users. (LT-LiDER: i3) For some of this hyper-personalised content, it is inevitable that NMT and LLMs will be used, especially when the content shelf-life is short, and the level of risk connected with inaccurate translations is low. As more robust technology enables continuous localisation workflows and lights-out project management, deadlines are likely to reduce even further. However, LT-LiDER interviewees 164 Dragoș Ciobanu <?page no="165"?> warned against simply accepting new realities without standing up for working conditions which safeguard quality results: In translation, there must be a resistance to working under extreme time pressure, as this affects the process and quality. Even if tools are used, time is necessary to do the job well. Read, evaluate, repair the result of the tool. We have to reclaim time. (LT-LiDER: i25) This will be particularly important because, “given that they may well be the only humans in an otherwise automated environment, translators will be expected to offer increased professional responsibility and liability.” (do Carmo and Moorkens 2022: 22). In our experiment, we identified participants who were able to gain more thinking time in the sound condition as a consequence of having to slow down to listen to the source and target segments instead of proceeding at their usual fast reading pace, comparing fragments of source and target segments in a dizzying back-and-forth motion. Nevertheless, their overall task time did not suffer, because the presence of sound also reduced the number of self-revision passes they performed - as participant 7 stated: “Listening to the target text sound made me feel more content with my work, as in ‘Yes, I can approve of this! ’”. In other cases, being able to listen to the source segments was an effective way of ensuring that the post-editors actually paid sufficient attention to the source segments instead of over-relying on the quality of the MT output: “If I am post-editing, I am focussing on the machine translated text, and correct or change it according to the needs. I am checking on the original segments only if the machine translated text is not really clear” (participant 16). As the quality of NMT and LLM output continues to improve, and errors become even more difficult to spot, having an audio version of the source and target text will supplement the translator’s usual visual content processing mode with additional information, helping language specialists deliver high-quality output. At the same time, there will be situations in which, although efficient workflows using MT may have been documented, it may be better for the project objectives to trust and allow translators to work differently: researchers may identify gains in terms of actual efficiency, but communication effec‐ tiveness is determined by the human interpretation of a dynamic communication act. […] to maximise what they interpret as being the optimal effect of their translations, translators often need to feel they are writing their own words, not maximising the use of MT.” (do Carmo and Moorkens 2022) So much technology, so little time 165 <?page no="166"?> 5 Conclusion To support technology-savvy language specialists to thrive and achieve a more ethical adoption of a range of technologies, including AI, in the language industry, a multifaceted approach is required, one that balances technological innovation with the preservation of human expertise and fair working conditions. First, transparency must become a cornerstone of technological inte‐ gration. Language service providers (LSPs) and technology developers should openly communicate the capabilities and limitations of their tools, ensuring that clients and end-users understand the role of automation in the final product. This includes disclosing when MT or LLMs are used, and clarifying for the benefit of clients and language specialists themselves the extent of human involvement in the final translated product. Second, ethical adoption necessitates a shift from an AI-first to a people-first mindset. This means designing tools and workflows that empower language professionals rather than replace them. Developers must actively involve trans‐ lators in the creation and refinement of new technologies, ensuring that these address real-world challenges without undermining the expertise of human specialists. For instance, tools should be ergonomic, adaptable, and capable of supporting the collaborative and nuanced decision-making required in translation tasks, rather than imposing rigid workflows that prioritise speed over quality. Third, the industry must continue to advocate for fair compensation mo‐ dels that reflect the complexity of tasks performed by language specialists. Metrics like edit distance or productivity rates, while partially useful for project management, should not be the sole determinants of remuneration. Instead, compensation should account for the cognitive effort, contextual analysis, and creative problem-solving inherent in translation work. This would help counteract the downward pricing trends that have made the profession less sustainable in recent years. Finally, fostering a culture of continuous learning and adaptability is essential. As technology evolves, so too must the skills of language professionals. Industry stakeholders, including LSPs, academic institutions, and professional associa‐ tions, should collaborate to provide accessible training programs that equip translators with the knowledge needed to navigate new technologies effectively. This proactive approach will not only enhance the resilience of the workforce, but also ensure that the human element remains central to the industry’s future. At the same time, similar collaborative initiatives should be directed at educating clients (alongside tomorrow’s clients now studying courses lacking Translation 166 Dragoș Ciobanu <?page no="167"?> and Interpreting Studies insights) about the limits of language technologies and the real value of language specialists. In conclusion, the language industry requires a more deliberate effort to pri‐ oritize human expertise, transparency, and fairness. By fostering collaboration between technology developers, language professionals, clients, and educators, the industry can harness the benefits of new technologies while safeguarding the integrity and sustainability of the profession. This balanced approach will not only address current challenges but also position the language services industry as a model for ethical innovation in the digital age. Bibliography ALPAC (1966). Language and Machines. Computers in Translation and Linguistics. A Report by the Automatic Language Processing Advisory Committee, Division of Behavioral Sciences, National Academy of Sciences, National Research Council. Automatic Language Processing Advisory Committee (ALPAC). Washington, D.C.: National Research Council. Bond, Esther (2018). Why People Leave the Language Industry. Slator. Available at: http s: / / slator.com/ why-people-leave-the-language-industry/ (Accessed: 11.05.2025) Briva-Iglesias, Vicent/ O’Brien, Sharon/ Cowan, Benjamin R. (2023). The Impact of Tradi‐ tional and Interactive Post-Editing on Machine Translation User Experience, Quality, and Productivity. Translation, Cognition & Behavior 6 (1), 60-86. Cahoon, Scott (2025). Re-Imagining Corporate Translations in the Era of AI. Accessible at: https: / / www.trados.com/ resources/ re-imagining-corporate-translations-in-the-er a-of-ai-at-dell/ Ciobanu, Dragoș (2016). Automatic Speech Recognition in the Professional Translation Process. Translation Spaces 5 (1), 124-44. Ciobanu, Dragoş/ Ragni, Valentina/ Secară, Alina (2019). Speech Synthesis in the Trans‐ lation Revision Process: Evidence from Error Analysis, Questionnaire, and Eye-Tra‐ cking. Informatics 6 (4), 51. Ciobanu, Dragoș/ Rios, Miguel/ Secară, Alina/ Brockmann, Justus/ Chereji, Raluca-Ma‐ ria/ Wiesinger, Claudia (2024). The Impact of Using Text-to-Speech (TTS) in Post-Edi‐ ting Machine Translation (PEMT) Workflows on Translators’ Cognitive Effort, Pro‐ ductivity, Quality, and Perceptions. Tradumàtica: Tecnologies de la Traducció 22, 323-354. do Carmo, Félix/ Moorkens, Joss (2022). Translation’s New High-Tech Clothes. In: Moor‐ kens, Joss/ De Luca, Elisabetta (Hrsg.). The Human Translator in the 2020s. London: Routledge, 87-104. So much technology, so little time 167 <?page no="168"?> ELIS (2025). European Language Industry Survey 2025: Trends, Expectations and Con‐ cerns of the European Language Industry. Available at: https: / / elis-survey.org/ wp-co ntent/ uploads/ 2025/ 03/ ELIS-2025_Report.pdf (Accessed 20.03.2026) Gaspari, Federico/ Toral, Antonio/ Naskar, Sudip Kumar/ Groves, Declan/ Way, Andy (2014). Perception vs. Reality: Measuring Machine Translation Post-Editing Produc‐ tivity. In: O’Brien, Sharon/ Simard, Michel/ Specia, Lucia (Eds.). Proceedings of the 11th Conference of the Association for Machine Translation in the Americas. Vancouver: Association for Machine Translation in the Americas, 60-72. Guerberof Arenas, Ana (2014). Correlations between Productivity and Quality When Post-Editing in a Professional Context. Machine Translation 28, 165-86. Inbox Translation/ Institute of Translation and Interpreting (ITI) (2023). Freelance Trans‐ lator Survey 2023. Accessible at: https: / / inboxtranslation.com/ resources/ research/ fre elance-translator-survey-2023/ (Accessed: 20.03.2026) Lambert, Joseph/ Walker, Callum (2024). Thriving or Surviving: Motivation, Satisfaction, and Existential Sustainability in the Translation Profession. Mikael: Kääntämisen ja Tulkkauksen Tutkimuksen Aikakauslehti 17 (1), 89-104. Lommel, Arle (2025). Navigating the Post-Localization Landscape: Trends Shaping the Future of Language Service. Presented at the ELEVATE 2025 online event. Lommel, Arle/ Uszkoreit, Hans/ Burchardt, Aljoscha (2014). Multidimensional Quality Metrics (MQM): A Framework for Declaring and Describing Translation Quality Metrics. Revista Tradumàtica: Tecnologies de la Traducció 12, 455-463. Matecat (n.d). How Does Matecat Calculate Payable Words? Available at: https: / / guides .matecat.com/ how-matecat-calculates-payable-words (Accessed: 11.05.2025) Moorkens, Joss (2020a). “A Tiny Cog in a Large Machine”: Digital Taylorism in the Translation Industry. Translation Spaces 9 (1), 12-34. Moorkens, Joss (2020b). Comparative Satisfaction among Freelance and Directly-Em‐ ployed Irish-Language Translators. Translation & Interpreting 12 (1), 55-73. Sakamoto, Akiko/ van Laar, Darren/ Moorkens, Joss/ do Carmo, Félix (2024). Measuring translators’ quality of working life and their career motivation: Conceptual and methodological aspects. Translation Spaces 13 (1), 54-77. Secară, Alina/ Rivas Ginel, Isabel/ Toral, Antonio/ Guerberof, Ana/ Ciobanu, Dra‐ goş/ Brockmann, Justus/ Plieseis, Claudia/ Raluca-Maria Chereji/ Rossi, Caroline (2025). LT-LiDER Language Technology Map - Technologies in Translation Practice and Their Impact on the Skills Needed. Available at: https: / / doi.org/ 10.25365/ phaidra.641 (Accessed: 20.03.2026) SFT (2022). Rapport de l’enquête 2022 Sur Les Pratiques Professionnelles En Traduc‐ tion. Available at: https: / / www.sft.fr/ docs/ 2024164729_sft-2022-enquete-traduction.p df (Access: 20.03.2026) 168 Dragoș Ciobanu <?page no="169"?> Terribile, Silvia (2023). Is Post-Editing Really Faster than Human Translation? . Transla‐ tion Spaces 12 (2), 1-29. Thompson, Brian/ Dhaliwal, Mehak Preet/ Frisch, Peter/ Domhan, Tobias/ Federico, Mar‐ cello (2024). A Shocking Amount of the Web Is Machine Translated: Insights from Multi-Way Parallelism. In: Ku, Lun-Wei/ Martins, Andre/ Srikumar, Vivek (Eds.). Fin‐ dings of the Association for Computational Linguistics: ACL 2024. Bangkok, Thailand: Association for Computational Linguistics, 1763-1775. Varga, Laszlo K./ Groot Nibbelink, Marjolein (2025). The 2025 Nimdzi 100 - The Size and State of the Language Services Industry. Available at: https: / / www.nimdzi.com/ nimd zi-100-2025/ (Accessed: 11.05.2025) Vieira, Lucas Nunes (2020). Automation Anxiety and Translators. Translation Studies 13 (1), 1-21. Yamada, Masaru (2023). Post-editing and a sustainable future for translators. In: Lacruz, Isabel (Hrsg.). Translation in Transition. Human and machine intelligence, Vol. XX. Amsterdam/ Philadelphia: John Benjamins Publishing Company, 39-56. So much technology, so little time 169 <?page no="171"?> 1 Die in diesem Beitrag verwendeten Analysen stammen aus einem Vortrag, den ich im Rahmen der von Nina Kalwa organisierten Vorlesungsreihe „Artificial Friday“ des Departments für Wissenschaftskommunikation des Karlsruher Institut für Technologie am 6. Mai 2022 gehalten habe. Ich bedanke mich für die Anregungen und Kommentare, die ich in diesem Zusammenhang von den Teilnehmer: innen erhalten habe und die hier eingeflossen sind. Wie tief tauchen Vermittelnde in den Großbereich „Künstliche Intelligenz“ ein? Vorschlag einer Messung 1 Jan Engberg Abstract: Das zentrale Anliegen dieses Beitrags besteht darin, zwei Bei‐ spiele für sogenanntes Content Marketing aus der Sicht des fachlichen Wissens zu beschreiben, das durch sie vermittelt wird. Als übergeordneter Rahmen für die Beschreibung zweier Angebote zum Wissensaufbau dient das systemtheoretisch geprägte WIKO-Modell (Budin 1996), das fachliches Wissen als gleichzeitig in stabile Strukturen und dynamische Prozesse or‐ ganisiert auffasst. Relevant ist hier, welche Rolle diese (fachliche) Organi‐ sierung des Wissens in den untersuchten Content Marketing-Texten spielt. Als spezifischeres Analysekonzept wird die Tiefe kausaler Erklärungen im Text (z. B. Engberg 2022) herangezogen. Auf dieser Grundlage werden zentrale Unterschiede bezüglich der Fachlichkeit des Wissens der beiden untersuchten Texte erfasst und dabei ermittelt, inwiefern das Content Marketing als Brücke zum fachlichen System aufgefasst werden können. 1 Das WIKO-Modell als Grundlage wissensbezogener Analysen Im Zentrum des vorliegenden Beitrags steht die Vermittlung fachlich-wissen‐ schaftlichen Wissens in Kontexten, die außerhalb der Wissenschaft liegen. In solchen Situationen ist es notwendig, die Komplexität nicht nur der Texte <?page no="172"?> an sich, also ihrer formalen Merkmale, sondern auch des zu vermittelnden Wissens im Vergleich zum wissenschaftlichen Ausgangspunkt zu reduzieren. Ein Kontext, in dem dies erfolgt, ist der des Content Marketings. Darunter verstehen wir mit Hilker (2017: 4) Folgendes: Content Marketing […] dient zur markenbezogenen Platzierung von Content über das Internet. Relevante Inhalte zeichnen sich durch einen inspirierenden, informativen, anregenden, unterhaltenden, emotionalen und teilbaren Charakter aus. Sie sind dabei vor allem eins: relevant und nicht werblich. Content Marketing unterstützt das Online-Marketing, das Branding und die Verkaufsförderung. (Hilker 2017: 4) Um die Vermarktung eines Produkts oder der eigenen Expertise zu unterstützen wird Wissen vermittelt. Dabei ist, anders als z. B. im Schulunterricht, die Wis‐ sensvermittlung kein Selbstzweck, sondern den Bedingungen des Marketings unterstellt. Mein Anliegen ist es zu zeigen, dass trotz des neuen Kontextes zen‐ trale Aspekte der kognitiv und systemtheoretisch orientierten Terminologiewis‐ senschaft einen wesentlichen Erklärungsbeitrag leisten können, insbesondere bei der Erfassung von inhaltlichen Komplexitätsgraden der Vermittlungstexte. Ausgangpunkt ist dabei eine Arbeit von Gerhard Budin, die einen wichtigen Beitrag zur Entwicklung kognitiver Ansätze in der Fachkommunikationsfor‐ schung geleistet hat (Budin 1996). In diesem Buch öffnet der Verfasser für die Terminologie und Fachkommunikationsforschung mit seinem Wissens-, Informations- und Kommunikations-Organisations-Modell (WIKO) ein Tor zur Denk-Welt der Systemtheorie mit ihrer Kombination von Stabilität durch vom Fach anerkannte (Wissens und Kommunikations-)Strukturen, auf der einen Seite, und von Dynamik durch die andauernden Ausgleichs- und Anpassungs‐ prozesse im Fach als Reaktion auf interner, autopoetischer Wissensanreiche‐ rung, auf der anderen. Budin (1996: 35) spricht hier von einer Dichotomie zwischen Zustand und Prozess. Der Großteil der theoretischen Ansätze insbe‐ sondere der Terminologiewissenschaft, die dieser Arbeit vorausgingen, haben das Augenmerk stärker auf Zustände (= Strukturen) als auf Prozesse gerich‐ tet. Die große Leistung der Arbeit lag und liegt darin zu zeigen, dass eine Beschreibung möglich ist, die beide Sichtweisen kombiniert und die folglich eine bessere Modellierung der tatsächlichen Merkmale fachlicher Kommunikation und Wissensausbau ermöglicht. Damit wurde ein maßgeblicher Beitrag zur Entwicklung der dritten Art von Modellierungsansätzen bei, aus denen die 172 Jan Engberg <?page no="173"?> 2 Für einen Überblick über den heutigen Stand dieser Art von Ansätzen, vgl. Engberg (2025). moderne Fachkommunikationsforschung nach Roelcke (2025: 14-18) besteht, und zwar die kognitiv-linguistische Modellierung. 2 Zentral im Ansatz von Budin steht die folgende zusammenfassende Aussage: Wissen ohne Kommunikation bleibt „Privatsache“, Information bleibt ohne Kommu‐ nikation „unentdeckt“, Kommunikation ohne Wissen „ohne Sinn“. Ohne Terminologie (also Strukturen von Begriffen und deren Repräsentationen) als Organisationsprinzip wäre Wissen, Information und Kommunikation in wissenschaftlichen und fachlichen Handlungsbereichen eine amorphe Masse ohne Ordnungsmuster und Prozeßökono‐ mie. (Budin 1996: 186) Im Fokus befindet sich hier die Idee, dass ein Kommunizieren über (fachliches) Wissen notwendig ist, damit es geteilt werden kann, dass aber gleichzeitig auch eine sprachlich (hier über Fachtermini) vermittelte Organisation des Wissens unerlässlich ist, damit dieses Wissen detailliert und interpersonell verwendbar gemacht werden kann. Wie ich diese Idee verstehe, handelt es sich dabei darum, mithilfe der fachintern akzeptierten Termini ein gemeinsames Wissensgerüst zu haben, über deren Struktur und Inhalt Einigkeit und damit Stabilität herrscht, wenn diese auch nicht unzerrütbar ist. Trotzdem muss man, damit man innerhalb eines wissenschaftlichen Fachbereichs Einsichten entwickeln kann, wissen, auf welcher Grundlage die Entwicklung stattfindet. Und hierzu dienen u. a. die fachlich definierten Begriffe und ihre zugehörigen sprachlichen Repräsentationen. Das Modell von Budin (1996) fokussiert insbesondere auf Prozesse und Zustände, die zur Bildung, Stabilisierung und Entwicklung wissenschaftlicher Begriffe und ihrer Bezeichnung beitragen bzw. ihre Voraussetzung bilden. Somit steht die fachinterne Kommunikation im Mittelpunkt. Die Grundprinzipien des WIKO-Modells lassen sich aber nach meiner hier zu begründenden Auffassung auch verwenden, um zu beschreiben und zu modellieren, wie mit Wissen in Situationen umgegangen wird, in denen das wissenschaftliche Wissen nicht fachintern aufrechterhalten oder entwickelt, sondern als Element in fachexter‐ ner Kommunikation eingeführt wird. Wie oben angeführt möchte ich hier als Beispiel die Vermittlung von Wissen zu Werbezwecken (genauer: in der Form des Content Managements, vgl. 1.3 unten) in den Mittelpunkt stellen. Wie aus dem obigen Zitat aus Budin (1996) hervorgeht, ist nach dem WIKO-Modell der Aspekt der Organisation von Wissen und Information zentral für die fachliche Kommunikation, da er ihr definierendes Merkmal ausmacht. Wie tief tauchen Vermittelnde in den Großbereich „Künstliche Intelligenz“ ein? 173 <?page no="174"?> Daraus folgt u. a., dass eine fachexterne Vermittlung zu Werbezwecken diese Organisation kommunizieren muss, damit die Fachlichkeit als Werbeargument Geltung erhalten kann. Im Anschluss hieran soll in meinem Beitrag anhand von zwei exemplarischen Analysen der Frage nachgegangen werden, wie unter‐ schiedliche Typen der werbebezogenen Wissensvermittlung aus der Perspektive der vermittelten Organisation des Wissens unterschieden werden können. Der Fokus soll dabei auf der Tiefe der angebotenen kausalen Erklärungsrelationen liegen. Im Folgenden wird zuerst der gewählte thematische Bereich (Künstliche Intelligenz) als Gegenstand der Untersuchung erläutert. Daraufhin soll die Perspektive der Wissenskommunikation in ihren Relationen zum WIKO-Modell erörtert und anschließend das Analyseinstrumentarium der Erklärungstiefe eingeführt werden. Daraufhin stelle ich zwei Beispielanalysen vor, deren Ergeb‐ nisse dann abschließend perspektiviert werden sollen. 2 Grundlagen 2.1 Wissenskommunikation als Rahmen Die hiesigen Ausführungen sind auf der Grundlage des Ansatzes der Wissens‐ kommunikationsforschung entstanden, der mit Inspiration von systemtheore‐ tischen Ansätzen wie die des WIKO-Modells entwickelt worden ist: The idea was and is to develop an open framework built on the following three pillars: expert knowledge is seen as constructed through communication, it is empirically present ways that reflect conditions of human cognition and it is expressed in a broad range of forms and formats that are textual in a broad sense and contextually situated. (Engberg et al. 2024: 1, meine Hervorhebung) Darüber hinaus, dass in dem Ansatz das Augenmerk auf Texte im weiten Sinne liegt, ist die Idee zentral, dass Wissen auf der Grundlage des kommu‐ nikativen Inputs (constructed through communication) in den Köpfen von Kommunikationsteilnehmer: innen durch Topdown- und Bottomup-Prozesse (reflect conditions of human cognition) konstruiert wird. Dies heißt, dass davon ausgegangen wird, dass bei der Kommunikation Texte als Grundlage und Input für Verstehensprozesse bei den Adressat: innen dienen, in denen Wissensstrukturen aufgebaut werden. Dabei ist die Annahme, dass Texte nur mittelbar den Aufbau der Wissensstrukturen bedingen, da menschliches Verstehen stark topdownorientiert auf der Grundlage von Erwartungen erfolgt. Folglich kann über Textanalysen nicht erforscht werden, wie Texte konkret verstanden werden, sondern lediglich, zum Aufbau welcher Wissensstrukturen 174 Jan Engberg <?page no="175"?> sie potenziell Anlass geben. Im Rahmen des WIKO-Modells heißt dies, dass untersucht werden kann, welche Organisation des Wissens durch die Texte und der darin beschriebenen Begriffe und Relationen angeboten wird, um das prinzipiell amorphe Wissen bei Adressat: innen fachlich zu strukturieren. Solche Angebote möchte ich unten in den Beispielanalysen erfassen. 2.2 Künstliche Intelligenz als Gegenstand fachexterner Vermittlung Der hier bewusst unscharf gehaltene Inhaltsbereich der Künstlichen Intelligenz ist in unserem Alltag allgegenwärtig, vom Handy über die Arbeitswelt hin zur Überwachung unterschiedlicher Art, um nur einige Bereiche zu nennen. Ihre Bedeutung hat zur Folge, dass Anbieter: innen von Produkten und Dienstleis‐ tungen häufig auf Künstliche Intelligenz verweisen. Folglich muss entschieden werden, wievel Fachwissen in welcher Komplexität vermittelt werden soll. Hier soll der Fokus der hiesigen Untersuchung liegen. Der Marketing-Kontext wird deshalb als Rahmen verwendet, weil hieraus das Merkmal erwächst, dass die Wissensvermittlung kein Selbstzweck ist, sondern anderen Zielen untergeordnet ist. Die allermeisten Bürger: innen sitzen lediglich am Benutzerende der KI, haben ein funktionales Alltagswissen darüber und wissen wenig bis nichts über die technischen Hintergründe für ihre Funktion. KI gehört somit zu unserer alltäglichen Standardwelt: Standardwelten sind zu verstehen als gesellschaftliche Alltagswelten, von denen ich weiß, dass auch andere nicht an ihrer objektiven Realität zweifeln, es sind die Bezugswelten, deren Elemente gesellschaftlich als fraglos existierend betrachtet und behandelt werden. Diese Welt des für alle selbstverständlichen Wissens ist sehr groß. Deswegen kenne ich davon auch nur einen sehr kleinen Ausschnitt, noch weniger ist mir wirklich vertraut. (Adamzik 2018: 102-103) Wenn für einen Gegenstand (hier: die KI) die alltägliche Standardwelt als Bezugswelt indiziert wird (z. B. durch Betonung seiner Alltäglichkeit), hat dies als Konsequenz, dass er als mithilfe alltäglicher Logik zu verstehen vorgestellt wird. Nach Adamzik ist ein Beispiel für Bezugswelten, die eine andere Logik beim Verstehen voraussetzen, religiöse und ideologische Welten (Adamzik 2018: 101). Durch die vermittlungsgetragene Verortung eines Gegenstandes in der alltäglichen Standardwelt wird also angezeigt, dass Adressat: innen beim Verstehen ihren Ausgangspunkt in ihrer existierenden Weltauffassung nehmen können und diese lediglich durch Spezifizierungen anreichern müssen. Wie tief tauchen Vermittelnde in den Großbereich „Künstliche Intelligenz“ ein? 175 <?page no="176"?> Die erwähnte Logik entspricht der fachlichen Organisation von Wissensele‐ menten, die im Rahmen des WIKO-Modells als definitorisches Merkmal für die jeweilige Fachlichkeit gilt. Dabei ist es aber je nach Zweck der Kommunikation (hier: Content Marketing) und angenommenem Interesse der Adressat: innen unterschiedlich, wie stark und wie umfassend die im Ausgangspunkt für Adressat: innen wahrscheinlich recht amorphe kommunizierte Wissensmasse in eine strukturierte Ordnung durch die Kommunikation gebracht werden soll: Die Textualisierung kann sich z. B. lediglich auf die Nennung des Begriffs in Form der terminologisierten Bezeichnung beschränken (wobei dann lediglich die Fachqualität aber nicht die fachliche Strukturierung angezeigt wird), oder Textverfasser: innen können mehr oder weniger informative Erklärungen in ihre Kommunikation einbauen (und dadurch versuchen, eine bestimmte fachliche Struktur in das potentiell amorphe Wissen zu bringen). Je stärker über die reine Nennung hinausgegangen wird, je mehr kann von einer Vermittlungs- und sogar komplexeren Popularisierungsabsicht gesprochen werden (Engberg et al. 2018). In den Beispielanalysen in 1.3 soll erfasst werden, wie komplex das Wissen über KI ist, das auf der Grundlage der untersuchten Texte von Adressat: innen aufgebaut werden kann. Dafür soll der Beschreibungsansatz der Tiefe kausaler Erklärungen dienen, der im nächsten Abschnitt vorgestellt wird. 2.3 Tiefe kausaler Erklärungen als Analyseansatz Um Unterschiede zwischen Texten aus der Perspektive der Komplexität von Wissensstrukturen analytisch zu erfassen, die auf ihrer Grundlage potenziell aufgebaut werden können, habe ich in früheren Arbeiten ein Analyseinst‐ rumentarium vorgeschlagen (vgl. z. B. Engberg 2022). Grundlage ist das in der Psychologie entwickelte Konzept der Erklärungstiefe, bekannt als die Illusion of Explanatory Depth (Rozenblit und Keil 2002). Dieses als „Illusion“ bezeichnete Phänomen wurde empirisch nachgewiesen und beschreibt die Tatsache, dass Menschen im Allgemeinen in kommunikativen Interaktionen den Detailierungsgrad ihres Erklärungswissens über Konzepte systematisch überschätzen. Beispielsweise haben Rozenblit und Keil (2002) nachgewiesen, dass Studierende ohne Expertise im Bereich der Luftfahrt, bei konkreter Nach‐ frage und ohne spezifischen kommunikativen Kontext dazu tendieren, ihr Erklärungswissen über die Funktionsweise eines Hubschraubers als adäquat zu beurteilen, insbesondere wenn ihnen ein Bild eines Hubschraubers vorliegt. Wenn sie aber nach dieser Bewertung anschließend gebeten werden, die tatsächlichen Ursachen zu benennen, dass ein Hubschrauber schweben und senkrecht abheben kann, sind sie regelmäßig nicht dazu in der Lage. Es lässt sich 176 Jan Engberg <?page no="177"?> 3 Hier und im Folgenden werden Originalzitate aus den beiden analysierten Texten angeführt. Für die Zwecke dieser Analyse spielt es keine Rolle, inwiefern die Aussagen dem jetzigen oder damaligen fachlichen Wissensstand entsprechen. Wichtig ist, dass die Sender: innen gewählt haben, Wissen mit einer bestimmten Strukturierung und Komplexität zu vermitteln. daraus ableiten, dass die Wissensstrukturen der Proband: innen, die nicht auf eigene Erfahrungen oder auf spezifisches wissenschaftliches Wissen beruhen, eine skelettartige und fragmentarische Beschaffenheit aufweisen, die anders als ähnliche Expertisestrukturen nicht ausreichend detailliert sind, um die Grund‐ lage tiefergehender kausaler Erklärungen zu bilden. Die Wissensstrukturen erfüllen aber die unmittelbar vorgestellten Anforderungen an die Erklärung in Nicht-Expertise-Situationen, sodass die Proband: innen mit ihrem Wissen funktional zufrieden sind. Hier sehen wir eine Konsequenz aus der oben vorgestellten Idee nach Adamzik (2018), nach der der Bereich der Technik als Teil unserer alltäglichen Standardwelt aufgefasst wird. Folglich gehen wir auch grundsätzlich davon aus, dass wir die Technik verstehen. Erst wenn wir in eine kommunikative Situation versetzt werden, in der kontextbedingt ein höheres Maß an Detailwissen erforderlich ist (z. B. durch detailliertere Fragen in einem experimentellen Umfeld), wird uns evident, dass die Komplexität unseres Wissens doch nicht ausreicht, um tiefere kausale Erklärungen aufzustellen. Nichtexpert: innen sind als Ausgangspunkt mit ihren skelettartigen Darstellun‐ gen von Konzepten zufrieden, bis kommunikative Anforderungen im Kontext sie zum Umdenken zwingen. Auf der Grundlage der Untersuchung der Illusion of Explanatory Depth haben die Forscher eine Skala verschiedener Grade der Erklärungstiefe vorgelegt, die mir bei meinem Vorhaben als Grundlage dient, um den Inhalt von Äußerungen in Vermittlungstexten, bei denen ich Erklärungsqualität interpretiere, qualitativ zu beschreiben. Die Psychologen unterscheiden zwischen drei Typen von Skelettstrukturen des Erklärungswissens: Causal relevance, Causal powers und Causal relations (Keil 2003: 675-680; Keil 2011: 254-255). Um die Besonderheit der letztgenannten Kategorie zu unterstreichen, ist ihre Bezeichnung im vorlie‐ genden Ansatz geändert und wird stattdessen als kausales System bezeichnet. Darüber hinaus ist es für die Analyse von Fällen wie dem hier im Fokus stehenden (Verbreitung von wissenschaftlichem Wissen zu Werbezwecken) relevant, eine vierte Kategorie zu ergänzen, die aus Studien über Strategien der Wissensvermittlung bekannt und hierarchisch vor dem Typus kausale Relevanz angesiedelt ist, nämlich die Denomination (vgl. z. B. Calsamiglia und Van Dijk 2004). Die vier Typen lassen sich wie folgt beschreiben: 3 Wie tief tauchen Vermittelnde in den Großbereich „Künstliche Intelligenz“ ein? 177 <?page no="178"?> - Denomination: Die Elemente werden beschrieben, ohne dass kausale Elemente isolierbar sind: „Ein Algorithmus besteht in einer Folge von eindeutigen Anweisungen.“ (Scopevisio 2017). Da bei solchen Beschrei‐ bungen häufig auf einen übergeordneten Begriff Bezug genommen und der beschriebene Begriff somit klassifiziert wird, können schwache Kau‐ salrelationen aus diesem übergeordneten Begriff mitschwingen, die in dem Begriff zentral angelegt sind. So bedeutet die Klassifikation eines Algorithmus als Folge von Anweisungen, dass sie als nutzenorientiertes Mittel gekennzeichnet wird (sie weist an), weshalb ihr irgendeinen Nutzen für menschliche Aktivität zugeschrieben wird. Auf diese Kausalrelation wird aber nicht kommunikativ fokussiert. - Kausale Relevanz: Elemente werden als miteinander kausal verbunden dargestellt, aber ohne Kausalrichtung. Der Fokus liegt auf die Bedeutsam‐ keit. Beispiel: „Sprachassistenten wie telegra KIT funktionieren sehr menschlich.“ (Telegra 2018). Die Sprachassistenten befinden sich in einer funktionalen Relation mit der Welt, und zwar eine menschenähnliche. Aber welche wird nicht vertieft. - Kausale Kraft: Die dargestellte kausale Relation vermittelt Wissen über die kausale Rolle eines Elements in Bezug auf ein anderes für das Funktio‐ nieren eines Objekts. Beispiel: „Die eigentliche Leistung des Algorithmus besteht darin, die Intention des Anrufers zu erkennen.“ (Telegra 2018). Der Algorithmus sorgt für eine Erkennung der Intention, wirkt damit auf diese Erkennung. - Kausales System: Die dargestellte kausale Relation hat Systemcharakter, mit kausalen Strukturen auf mehreren Ebenen. Beispiel: „Systeme wie der telegra KIT beherrschen durch Nutzung neuronaler Netze […] die Er‐ kennung von Informationsinhalten durch die Kontextanalyse.“ (Telegra 2018) Für alle vier Typen von erklärenden Textstrukturen gilt, dass sie in der Weise grob sind, dass aus ihnen ohne fachliches Vorwissen keine Wissensstrukturen aufgebaut werden können, auf deren Grundlage Erklärungen auf Expertise‐ ebene erstellt werden können. Um ein vermittlungsbezogenes Verständnis zu erreichen, das für Adressat: innen in bestimmten Kommunikationssituationen zufriedenstellend ist, können sie ausreichend sein. Dabei zeigt die Wahl an Komplexität der Erklärungsstrukturen auch, welches Verstehensniveau bei Adressat: innen angestrebt wird und somit auch, wo der Fokus der Vermittlung liegt. Wird eine Einsicht z. B. in die Funktionsweise der KI oder eines Algorith‐ mus angestrebt, oder soll lediglich ein Produkt durch die Erwähnung davon gekennzeichnet (und aufgewertet? ) werden, dass KI bei ihm eine Rolle spielt? 178 Jan Engberg <?page no="179"?> 4 Zur Form der Beispiele: Die Beispiele sind im Wortlaut zitiert. Wo z. B. das Bezugs‐ element eines Pronomens fehlt, wird dieses in eckigen Klammern eingefügt. Runde Dieser Frage möchte ich in der folgenden Analyse unter Anwendung dieses Instrumentariums nachgehen. 3 Analyse Die bisherigen Überlegungen sollen in diesem Abschnitt durch zwei exem‐ plarische Analysen von Texten konkretisiert werden, in denen über den In‐ haltsbereich der KI im weiten Sinne gesprochen wird, aber ohne, dass ein tieferes Wissen vorausgesetzt wird, um den angepeilten Kommunikationszweck zu realisieren. In beiden Fällen handelt es sich um Texte, die dem Content Marketing (vgl. 1.1) zugeschrieben werden können. Bei der Vermittlung wird Inhalt vorgestellt, der die Vermarktung von Produkten oder Eigenexpertise des Unternehmens unterstützen kann, ohne selber Werbung im engeren Sinne zu sein. Aus Vermittlungsperspektive sprechen wir hier konkret von Texten, die von der KI handeln, aber nicht den Aufbau einer detaillierteren Wissensstruktur bei den Adressat: innen anstreben. In den Worten des Zitats in 1.1 oben aus Budin (1996) soll die Amorphität des Wissens der Adressat: innen über KI durch Ordnungsmuster organisiert werden, aber nur zu einer Erklärungstiefe, die dem Kommunikationszweck der Sender: innen entspricht. Die zu erforschende Analysefrage lautet deshalb: Zur Bildung welcher Wissensstruktur mit welcher Erklärungskomplexität wird in dem untersuchten Text angeregt? 3.1 Situation: Erklärung der Funktion Beim ersten Beispiel handelt es sich um einen BlogBeitrag aus dem Jahre 2018 mit dem Titel KI im telefonischen Kundenservice: Was Voicebots heute schon leisten (Telegra 2018). Der Blog gehört zur Website des Call Center-Anbieters telegra, und der untersuchte Beitrag ist im Zusammenhang mit der Vermarktung des sogenannten Voicebots des Anbieters zu sehen. Dabei wird darauf Wert gelegt, dass der Voicebot eine KI-gestützte Einheit ist und dass er bestimmte Aufgaben lösen kann. Im Folgenden liegt das Augenmerk lediglich auf Aussa‐ gen, die der KI als Grundlagentechnologie betreffen. Aussagen dazu, was der Voicebot mithilfe der KI kann, sind dagegen nicht berücksichtigt worden, da hier ja die Content Marketing-Funktion interessiert. Zu der KI werden in dem Blogbeitrag die folgenden Textelemente angeboten, die ein Element kausaler Erklärung beinhalten: 4 Wie tief tauchen Vermittelnde in den Großbereich „Künstliche Intelligenz“ ein? 179 <?page no="180"?> Klammern entsprechen dem Originaltext. Fettdruck stammt von mir und gibt die Elemente an, die für meine Interpretation ausschlaggebend sind. (1) Im Handumdrehen sorge sie [Googles Künstliche Intelligenz] in einem Ge‐ spräch zwischen Mensch und Maschine für eine Terminvereinbarung. (2) Das eigentlich erstaunliche an Google Duplex ist nicht die Tatsache, dass ein Algorithmus es schafft, den Prozess einer Terminvereinbarung abzubilden. (3) Sie [eine Maschine] muss erkennen, ob die Kommunikation noch in eine Richtung läuft, die am Ende zu einem Ergebnis führt. (4) Die Leistung der KI besteht nicht darin, vollautomatisiert die Antwort zu liefern. Sie besteht nicht mal darin, dass die Antwort sich anhört wie eine menschliche Stimme. Die eigentliche Leistung des Algorithmus besteht darin, die Intention des Anrufers zu erkennen (was wurde gewollt). Und genau das leisten Sprachassistenten wie telegra KIT heute bereits auf Mausklick. (5) Bereits heute sorgt KI dafür, dass einfache und wiederkehrende Anfragen (FAQs), erkannt und sinnbringend beantwortet werden können. Bevor sich dieser Beitrag der hier zentral gestellten kausalen Erklärungsrelati‐ onen in der aufzubauenden Wissensstruktur widmet, soll darauf hingewiesen werden, dass u. a. über Verweisketten eine Vereinfachung dadurch angestrebt wird, das KI mit Chatbot gleichgesetzt wird. So ist die Formulierung Googles Künstliche Intelligenz eine Wiederaufnahme von Google Duplex, was ein frühes Voicebot war. Dies ist natürlich eine starke Vereinfachung aus der Sicht des fachlichen Wissens (KI = Verwendung, die auf KI aufbaut), aber eben ein Beispiel für eine gewollte Vereinfachung, wahrscheinlich um den Fokus auf relevantere Aspekte des fachlichen Wissens aus der Sicht des Marketing-Ziels legen zu können. Aus Platzgründen wird dieser Aspekt hier aber nicht weiter ausgeführt. Die Äußerungen, die den Aufbau einer Wissensstruktur mit kausalen Rela‐ tionen nahelegen, befassen sich ausschließlich mit der Ebene kausaler Kräfte: Es wird angeführt, welche Effekte durch die KI erzielt werden können. Dagegen wird kein Wert daraufgelegt, bei Adressat: innen den Aufbau eines geordneten Wissens anzuregen, auf dessen Grundlage das Funktionieren erklärt werden könnte. Interessanterweise ist das einzige Beispiel für eine Äußerung, die die Erklä‐ rungstiefe kausaler Systeme erreicht, der folgende Textauszug: (6) Systeme wie der Voicebot telegra KIT beherrschen durch Nutzung neuronaler Netze (und zugänglicher Technologie von z. B. Google) die Erkennung von Informationsinhalten durch die Kontextanalyse. 180 Jan Engberg <?page no="181"?> Hier wird nicht nur eine kausale Kraft-Erklärung angeboten (Systeme Erken‐ nung), sondern es wird auch das bedingende System vorgestellt (neuronale Netze). Interessanterweise handelt es sich hier wohl um eines der technischen Kernstücke einer KI-Anwendung - aber dieser Zusammenhang wird nicht weiter thematisiert und gehört damit nicht zu der Ordnung des amorphen Wissens über KI, die den Adressat: innen angeboten wird. KI wird somit hier lediglich aus der Sicht ihrer Funktion und nicht ihres Funktionierens dargestellt. Man kann vielleicht sagen, dass hier KI in ihrer magischen Natur dargestellt wird, wobei die primäre Aussage ist, dass das Produkt des Anbieters eine moderne Technik mit funktionalen Besonderheiten enthält. 3.2 Situation: Erklärung des Funktionierens Auch im zweiten Fall wurde ein Blogbeitrag als Untersuchungsbeispiel ge‐ wählt (Scopevisio 2017). Die Website, auf der der Blog platziert ist, gehört zu einem Software-Unternehmen, das sogenannte Enterprise Resource Planing‐ Programme entwickelt und anbietet, d. h., Programme, mit denen Geschäfts‐ prozesse zentral gesteuert werden können. Es handelt sich also auch hier um Information über Technik, die für das von einem Unternehmen angebotene Produkt eine zentrale Rolle spielt. Aus der Sicht des für die Konstruktion ange‐ botenen Wissens wird hier aber eine etwas andere Strategie verfolgt. Auch hier ist die Analyse auf Aussagen zur KI beschränkt, die kausale Erklärungselemente enthalten. Auch in diesem Fall finden sich Beispiele für Aussagen, die Wissensaufbau auf der Ebene kausaler Kräfte zulassen: (7) So soll künstliche Intelligenz in den nächsten Jahren zunehmend die Unterneh‐ men und die Arbeitswelt revolutionieren. (8) Alexa, Siri, Cortana - die Chatbots von Amazon, Apple und Microsoft sind Bei‐ spiele dafür, wie künstliche Intelligenz die Interaktion zwischen Mensch und Computer einfacher und effizienter macht. Angeführt werden Ergebnisse, die als durch die KI bedingt indiziert werden. Die erklärungsbezogene Komplexität ist aber in diesem Blog generell höher als in der vorher beschriebenen Situation. Dies ist teils durch komplexere Ketten von Äußerungen auf der Ebene kausaler Kräfte, teils durch systembezogene Äußerungen ersichtlich. Im folgenden Beispiel findet sich höhere Komplexität in Form von Reihungen: Wie tief tauchen Vermittelnde in den Großbereich „Künstliche Intelligenz“ ein? 181 <?page no="182"?> (9) Schematisch ist der Ablauf in der (kontrollierten) Trainingsphase wie folgt: • Beispiel-Daten werden dem Netz als Input übergeben. • Der Output für die Beispiele ist bekannt. • Der tatsächliche Output wird mit dem Soll-Output verglichen. • Wenn beide voneinander abweichen, werden Schwellenwert und Gewichte zwischen den einzelnen Schichten so angepasst, dass der mittlere Fehler zwischen Soll und Ist sinkt. • Fehlerdaten werden durch das Netz zurückgegeben (Backpropagation). Hier wird eine Kette von Äußerungen aufgestellt, zwischen denen es in den meisten Fällen eine Verbindung auf der Ebene kausaler Kräfte gibt. Wir errei‐ chen inhaltlich nicht die Ebene kausaler Systeme, denn es werden keine Erklä‐ rungen der den kausalen Kräften zugrundeliegenden Systeme angeboten. Aber durch schematisch und Ablauf wird ein komplexer Zusammenhang zwischen den Elementen angezeigt. Zweitens gibt es hier auch Äußerungen, die die Grundlage für Wissenaufbau auf der Ebene kausaler Systeme bilden können. Ein erster Typ soll im folgenden Beispiel dargestellt werden. (10) Viele künstliche Intelligenzen arbeiten mit heuristischen Methoden. Diese Methoden sind geeignet, schnelle Lösungen zum Beispiel in Entschei‐ dungsbäumen zu finden. “Normale” Algorithmen brauchen vier Jahre, um einen Baum der Tiefe 15 und mit einer Verzweigungsrate von 3 zu durch‐ forsten. Heuristische Algorithmen schaffen das in Sekunden, zum Beispiel, indem sie den kürzesten Pfad zwischen zwei Knoten in einem Netzwerk finden. Dazu werden KI-Programme geschrieben, die auf Schätzungen und Beobachtungen basieren - auf so genannter Fuzzy Logic oder unscharfer Logik. In Beispiel 10 wird mit einer Aussage auf der Ebene kausaler Relevanz an‐ gefangen (Viele …). Im folgenden Satz erfolgt dann eine Spezifierung der kausalen Relation Richtung kausaler Kräfte, indem arbeiten mit als schnelle Lösungen in Entscheidungsbäumen zu finden spezifiziert wird. Später folgen dann Äußerungen, die eine Erweiterung des Wissensaufbaus auf die Ebene kausaler Systeme erlaubt, und zwar durch die Angaben der systemischen Grundlage für die schnellen Lösungen (kürzesten Pfad … finden, auf Schätzungen und Beobachtungen basieren … Fuzzy Logic). Adressat: innen können also auf der Grundlage dieses Textstücks potenziell die gelieferten Informationen so organisieren, dass sie wissen, welche Elemente zusammenarbeiten, welches Element die kausale Kraft ausführt und worauf diese kausale Kraft beruht. Eine zweite Art, Wissensaufbau auf der Ebene kausaler Systeme zu ermögli‐ chen, sehen wir in der folgenden Äußerung: 182 Jan Engberg <?page no="183"?> (11) Künstliche neuronale Netze sind parallel verbundene Netzwerke aus einfachen, adaptiven Elementen in hierarchischer Anordnung, die mit ihrer Umgebung in derselben Weise wie biologische Nervensysteme interagieren. Durch den metaphorischen Vergleich mit biologischen Nervensystemen erfolgt eine rumpfartige Systembeschreibung, wobei der Detaillierungsgrad der kausa‐ len Wissensstruktur natürlich stark von der Wissensgrundlage der jeweiligen Adressat: in abhängig ist. Die generell höhere Komplexität der erklärenden Äußerungen (komplexe Ketten, systembezogene kausale Erklärungen) in diesem Text ist auch durch die Einleitung zum Fazit des Blogs ersichtlich: (12) KI-Systeme arbeiten mit intelligenten Algorithmen. In diese Kategorie fallen heuristische Verfahren, künstliche neuronale Netze sowie evolu‐ tionäre (genetische) Algorithmen. Hier wird dargestellt, welche Elemente kausal an der KI beteiligt sind. Die Äußerung befindet sich bezüglich ihrer Komplexität zwar isoliert auf der recht oberflächlichen Ebene der kausalen Relevanz. Denn aus der Äußerung selbst kann lediglich interpretiert werden, dass die Elemente in einem kausalen Zusammenhang zur KI stehen. Jedes der drei Elemente sind aber zuvor so dargestellt worden, dass die gelieferten Informationen zu einem komplexeren Wissensaufbau auf der Ebene kausaler Systeme genutzt werden können. 4 Schlussbemerkungen Aus der Sicht des hier verwendeten qualitativen und interpretativen Analy‐ seinstrumentariums unterscheiden sich die beiden analysierten Texte in der Komplexität der Wissensstruktur, die auf ihrer Grundlage aufgebaut werden kann, auf dem Gebiet kausaler Erklärungen. In beiden Fällen handelt es sich wie angeführt um sogenanntes Content Marketing. Im ersten Fall liegt der Fokus deutlich auf der Rolle von KI für ein Produkt des Anbieters, während im zweiten Fall KI und ihre Merkmale im Mittelpunkt stehen; die Auswirkungen auf Produkte der Sender: innen werden dabei eher beiläufig erwähnt. Darauf aufbauend kann man sagen, dass im zweiten Fall eine umfassendere Struktu‐ rierung des amorphen Wissens über KI bei Adressat: innen als im ersten Fall angestrebt wird. Dadurch wird im zweiten Fall auch eine breitere Einführung der Adressat: innen in die fachliche Denke der KI-Technik angestrebt. Aus der Sicht der Bezugswelten nach Adamzik (2018) kann man sagen, dass im zweiten Fall Wie tief tauchen Vermittelnde in den Großbereich „Künstliche Intelligenz“ ein? 183 <?page no="184"?> 5 In anderem Zusammenhang habe ich diesen Komplex mit dem erweiterten analytischen Rahmen der gestuften Erklärungsambition und seinen Merkmalen erfasst, vgl. Engberg (2023). zum spezialisierten Technik-Teilbereich der Alltagswelt für die Adressat: innen ein Tor eröffnet werden soll, damit dieser Teilbereich umfänglich ein Teil des Alltags wird. 5 Im ersten Fall erfolgt dies auch, aber das Tor wird hier eher nur einen Spaltbreit geöffnet und es wird eher mit den „magischen“ Elementen der KI als Werbeargument gearbeitet. Im ersten Fall handelt es sich deshalb eher um Content Marketing mit Betonung des Marketings, wogegen im zweiten Fall die Betonung auf Content liegt, d. h. auf Wissensvermittlung. Die hier durchgeführte exemplarische Analyse sollte zeigen, dass die von Budin (1996) im WIKO-Modell angelegten Ideen und Prinzipien zur Rolle fachlicher sozialer Systeme bei der Strukturierung von Wissen, Information und Kommunikation auch außerhalb des engeren Bereichs fachinterner Kommuni‐ kation fruchtbar eingesetzt werden können. Dabei hat sich auch herausgestellt, dass dieser frühe Beitrag zum Einbezug systemtheoretischer Denkweisen in die Denke der Fachkommunikationsforschung nach wie vor relevante Einsichten ermöglicht, eine Tatsache, die die Bedeutung Gerhard Budins für die Entwick‐ lung unseres gemeinsamen Feldes, das der Fachkommunikationsforschung, nochmal unterstreicht. Bibliographie Adamzik, Kirsten (2018). Fachsprachen. Die Konstruktion von Welten. Tübingen: Narr Francke Attempto. Budin, Gerhard (1996). Wissensorganisation und Terminologie. Die Komplexität und Dynamik wissenschaftlicher Informations- und Kommunikationsprozesse. Tübingen: Narr. Calsamiglia, Helena/ Van Dijk, Teun A. (2004). Popularization Discourse and Knowledge about the Genome. Discourse & Society, 15, 369-389. Engberg, Jan (2022). Erklärungstiefe als Kennzeichen fachlicher Vermittlung. In: Heid‐ rich-Wilhelms, Franziska et al. (Hrsg.). Fachkommunikationswissenschaft gestern - heute - morgen. Hildesheim: Olms, 135-156. Engberg, Jan (2023). Dissemination of science and communicative efficiency of texts - is level of explanatory ambition a relevant diagnostic tool? trans-kom, 16, 4-21. Engberg, Jan (2025). Specialized Communication and cognition. In: Roelcke, Thorsten et al. (Hrsg.). Specialized Communication. Berlin/ Boston: de Gruyter, 67-86. Engberg, Jan/ Luttermann, Karin/ Cacchiani, Silvia/ Preite, Chiara (2018). Studying popu‐ larisation in legal communication: Introduction and overview. In: Engberg, Jan/ Lutter‐ 184 Jan Engberg <?page no="185"?> mann, Karin/ Cacchiani, Silvia/ Preite, Chiara (Hrsg.). Popularization and Knowledge Mediation in the Legal Field. Münster: Lit Verlag, IX-XXV. Engberg, Jan/ Fage-Butler, Antoinette/ Kastberg, Peter (2024). Introduction. In: Engberg, Jan/ Fage-Butler, Antoinette/ Kastberg, Peter (Hrsg.). Perspective on Knowledge Com‐ munication: Concepts and Settings. Routledge: New York, 1-16. Hilker, Claudia (2017). Content Marketing in der Praxis: Ein Leitfaden - Strategie, Konzepte und Praxisbeispiele für B2B- und B2C-Unternehmen. Wiesbaden: Springer Fachmedien. Keil, Frank C. (2003). Categorisation, causation, and the limits of understanding. Lang‐ uage and Cognitive Processes, 18, 663-692. Keil, Frank C. (2011). The Problem of Partial Understanding. In: Petersen, Margrethe/ Eng‐ berg, Jan (Hrsg.). Current Trends in LSP Research. Aims and Methods. Frankfurt a.M.: Peter Lang, 251-274. Roelcke, Thorsten (2025). Models and concepts of Specialized Communication. In: Roelcke, Thorsten/ Breeze, Ruth/ Engberg, Jan (Hrsg.). Specialized Communication. Berlin/ Boston: de Gruyter, 1-30. Rozenblit, Leonid/ Keil, Frank C. (2002). The misunderstood limits of folk science: an illusion of explanatory depth. Cognitive Science, 26, 521-562. Scopevisio (2017). Künstliche Intelligenz. Available at: https: / / www.scopevisio.com/ b log/ kuenstliche-intelligenz/ wie-funktioniert-ki-kuenstliche-intelligenz/ (Accessed: 20.03.2026) Telegra (2018). KI Kundenservice Telefo. Available at: https: / / blog.telegra.de/ ki-kunden service-telefo (Accessed: 20.03.2026) Wie tief tauchen Vermittelnde in den Großbereich „Künstliche Intelligenz“ ein? 185 <?page no="187"?> Text- und Diskursverkettung: Ein Beitrag zur Intertextualität Wolfgang Dressler Abstract: Dieser Beitrag schlägt das neue Konzept der Textbzw. Dis‐ kursverkettung vor. Dadurch werden abgeschlossene Texte zu Texten höherer Ordnung verkettet. Beispiele sind eine Heilige Messe und Bachs Weihnachtsoratorium. Die Verkettung erfolgt auf zwei Ebenen: Auf der ersten für jeden der sechs Teile des Weihnachtsoratoriums, indem beste‐ hende Texte mit neuen Texten in einer motivierten Anordnung verkettet werden. Auf der zweiten die Wiederverkettung der Teile zum gesamten Weihnachtsoratorium. Diese folgt dem ordo naturalis, da die Anordnung der Teile der Kalenderchronologie der Tage der von Bach intendierten Aufführungen folgt. Die Motivationen der (Wieder)verkettung ergeben sich aus Prinzipien und (nach der Natürlichen Linguistik) kognitiv und semiotisch begründeten Tendenzen der pragmatischen und semantischen Kohärenz, der Kohäsion und Situationalität: Pragmatische Identitäten sind wichtiger als syntaktische und lexikalische. Verkettungen sind immer stärker als Wiederkettungen. 1 Einleitung Dieser Beitrag zur Intertextualität von Texten und Diskursen führt die mit einander verwandten Begriffe Text- und Diskursverkettung, verkettete Texte bzw. Diskurse ein und versucht zu zeigen, dass diese Begriffe terminologiewis‐ senschaftlich adäquater als ähnliche Begriffsvorschläge sind, um fix verkettete und angeordnete Verbindungen von Texten bzw. Diskursen zu bezeichnen. Die ersten ausführlich analysierten Beispiele sind eine katholische Heilige Messe und Johann Sebastian Bachs Weihnachtsoratorium, welches aus geschriebenen, gesprochenen und gesungenen Texten besteht. Die Verkettung erfolgt dabei auf zwei Ebenen: Auf der ersten erfolgt die Verkettung für jeden der sechs Teile des Weihnachtsoratoriums, auf der zweiten die Verkettung der sechs <?page no="188"?> Teile zum gesamten Weihnachtsoratoriums. Diese folgt in Analogie zum ordo naturalis auf der einfachen Textebene. Die Anordnung der Verkettung ist immer entweder konventionell oder vom Autor selbst intendiert, während die Reihenfolge von Erstellung einer Übersetzung nach der Erstellung des Originals naturgegeben ist, ebenso die Reihenfolge Nacherzählung nach Erzäh‐ lung, Vorlesungsmitschrift nach Originaltext der Vorlesung. Eine kognitive und situative Voraussetzung erfolgreicher Textverkettung ist, dass Adressant: innen und intendierte Adressat: innen sowie die intendierte Situation der Performanz, d. h. der aktuellen Produktion und Rezeption, gleich bleiben. 2 Theoretischer Rahmen und kurze Forschungsgeschichte Den theoretischen Rahmen bildet die Natürliche Linguistik (ab jetzt NatLing), wie sie in Dressler (1989), Dressler und Kilani-Schoch (2016) und Dressler (2024) dargestellt wird, vgl. auch Dziubalska-Kołaczyk et al. (2025). Das Modell der NatLing ist semiotisch und kognitiv begründet und geht von daraus resultie‐ renden universell zur Verfügung stehenden Präferenzen aus. Für diesen Beitrag sind zwei Präferenzen relevant: 1) Die Präferenz für möglichst große Ikonizität, d. h. Analogie zwischen Inhalt und sprachlicher Form. 2) Die Präferenz der Indexikalität für Adjazenz und gegen große Distanz, z. B. zwischen einem Demonstrativpronomen und seinem Antezedens, auf welches es sich bezieht. Beaugrande und Dressler (1981) haben den schon vorher existierenden Be‐ griff der Intertextualität von Texten und Diskursen neu eingeführt und systema‐ tisch analysiert, z. B. von Kritiken, Kommentaren, Parodien, Briefwechsel, ohne etwas Spezifisches zur hier relevanten Text/ Diskursverkettung beizutragen (vgl. Adamzik 2004, 2016). Beispielsweise sind in einem in einer Zeitung fortlaufend erscheinenden Fortsetzungsroman die einzelnen Fortsetzungen miteinander verkettet, während ein Kommentar zu einer Fortsetzung mit dieser nur inter‐ textuell assoziiert ist. An den Textverkettung nur oberflächlich anklingende Konzepte wie Palimp‐ seste und Trans-, Meta- und Hypertextualität (Genette 1993), kooperative Intertextualität (Krause 2000), Textvernetzung (Adamzik 2011: 369), treffen nicht, was in diesem Beitrag zur Festschrift Budin mit Textverkettung intendiert ist, vgl. auch Gruber et al. (2006), Klein (1991), Reisigl (2011), Reisigl und Wodak (2001: 221). Verkettete Texte ergeben einen Supertext, eine Wiederverkettung dieser Su‐ pertexte einen Supertext höherer Ordnung, wie in Bachs Weihanctsoratorium. 188 Wolfgang Dressler <?page no="189"?> 3 Beispiele Ausführlich analysiert werden zwei Beispiele: A) Die heilige Messe besteht aus zwei Hauptteilen: der Liturgie des Wortes (Wort‐ gottesdienst) und der eucharistischen Liturgie (Eucharistiefeier). Diese beiden sind eng verbunden und bilden eine gottesdienstliche Einheit; sie werden von Eröffnungs- und Abschlussriten eingerahmt. Das Wort Messe (lateinisch Missa) leitet sich von der Entlassungs und Sendungsformel der lateinischen Liturgie Ite, missa est! (wörtlich Geht hin, es ist die Aussendung! ) ab. Das Wort missa bezeichnete im Spätlateinischen Entlassung, Verabschiedung Die-verketteten-Textteile-der-Messe-sind-numeriert-a),-b),-c)….: Eröffnung: - a) Einzug - Gesang: Zumeist singt die Gemeinde zur Eröffnung ein Lied (Introitus) (Kreuzzeichen) - b) Liturgischer-Gruß,-ggf.-Einführung - c) Allgemeines Schuldbekenntnis oder Taufgedächtnis, Misericordia (meist dreimal Erbarme Dich unser oder ähnlich) im Stufengebet - d)- Kyrie - e) Gloria - f) Tagesgebet (je nach Anlass): Wortgottesdienst (Liturgie des Wortes); Antwort Hallelujah ist Teil desselben Diskurses - g) Erste Lesung (Lesungen aus dem Alten und dem Neuen Testament je nach Anlass): Antwortpsalm ist Teil desselben Diskurses - h) Zweite-Lesung - i) an-Sonntagen-und-Festtagen-dritte-Lesung - j) Ruf-vor-dem-Evangelium - k) Evangelium - l) Homilie-(Predigt) - m) Glaubensbekenntnis-=-Credo - n) Fürbitten-(mehrere-Fürbitten-bilden-einen-einzigen-Listendiskurs) - o) Eucharistiefeier-(Eucharistische-Liturgie) (Gabenbereitung-für-Kommunion,-Bereitung-des-Altares,-Gabenprozes‐ sion) (Bei der Gabenbereitung werden der Altar bereitet und Hostienschale (Patene), Kelch, Brot, Wein und Wasser zum Altar gebracht), was meist von - p) Gesang,-dem-Offertorium,-begleitet-und-durch-das Text- und Diskursverkettung: Ein Beitrag zur Intertextualität 189 <?page no="190"?> r) Gabengebet-des-Priesters-abgeschlossen-wird. Gabengebet,-Fürbitten: - s) Eucharistisches Hochgebet (Herr, ich bin nicht würdig oder ähnlich), darin (sc. im selben Diskurs) Präfation mit Sanctus - t) Post-Sanctus, Anamnese mit Einsetzungsbericht, (Wandlung und Akkla‐ mation) Epiklese,-Interzessionen-und-Doxologie - u) Gebet-des-Herrn-(Vaterunser) - v) Friedensgruß - w) (Brechung-des-Brotes-mit)-Agnus-Dei-(Lamm-Gottes) - ex) Dankgesang - y) Schlussgebet-=-Dankgebet-(Postcommunio) Abschluss: -gegebenenfalls: - z) Vermeldungen Abgeschlossen mit dem Segen und der liturgischen Entlas‐ sung (Ite missa est, Gehet hin in Frieden).Zum Auszug des Priesters und sei‐ ner-Assistenz-wird-häufig-ein - za) Schlusslied-gesungen. B) - Johann-Sebastian-Bachs-Weihnachtsoratorium Zwei Ebenen der Textverkettung finden wir bei Bachs Weihnachtsoratorium: Eine erste Ebene betrifft die Verkettung verschiedenartiger Typen verschiedener Autoren jedes der sechs Teile. So verkettet der erste Teil a) den Chor Jauchzet, frohlocket! …, b) ein Rezitativ des Evangelisten (Bass) nach Lukas (2.1 und 3.6) des Neuen Testaments, c) das Alt-Rezitativ Nun wird mein liebster Bräutigam…, d) die Alt-Arie Bereite dich, Zion,…, e) Paul Gerhardts Choral Wie soll ich dich empfangen…, f) Martin Luthers Choral Er ist auf Erden kommen arm…, g) die Bass-Arie Großer Herr, oh starker König…, h) Luthers Choral Ach mein liebes Jesulein…. Die zweite Ebene betrifft die Verkettung der sechs Teile, die chronologisch durch die Reihenfolge der Feiertage bestimmt ist, an denen sie aufgeführt werden sollen, also: Teil I: Am ersten Weihnachtsfeiertage; Teil II: Am zweiten Weihnachtsfeiertage; Teil III: Am dritten Weihnachtsfeiertage; Teil IV: Am Feste der Beschneidung Christi; Teil V: Am Sonntage nach Neujahr; Teil VI: Am Feste der Beschneidung Christi. Die Intentionalität Johann Sebastian Bachs und die traditionsbedingte Ak‐ zeptabilität beim Leipziger Kirchenvolk haben bestimmt, welche Texte für jeden Teil verkettet worden sind, die intertextuelle Kohärenz (d. h. semantischer bzw. 190 Wolfgang Dressler <?page no="191"?> pragmatischer Zusammenhang) und besonders Kohäsion (d. h. syntaktischer bzw. morphologischer bzw. prosodischer Zusammenhang) innerhalb eines Teils ist stärker als zwischen den Teilen. So sind in Teil II die einzigen anaphorischen Verweise auf Teil I einerseits die Protagonisten Gott und Christus samt Engeln, andererseits im Eingangsrezitativ die anaphorische Präpositionalphrase in der‐ selben Gegend. In der zweiten Zeile des Eingangschors von Teil I bezieht sich Gesang auf das Singen im Schlusschoral von Teil 2. Derartige anaphorische Bindungen zwischen Anfang von Teil 4 und Ende von Teil 3 gibt es nicht, dasselbe gilt für die folgenden Teile. Es besteht also ein deutlicher Unterschied zwischen den ersten drei im Dezember aufzuführenden Teilen und den für Januar gedachten Teilen 4 bis 6. C) Gewöhnlich sind die Strophen eines Gedichts oder Lieds eigene autonome Texte, die verkettet worden sind. Nehmen wir als Beispiel die ersten drei der sechs Strophen von Goethes Schlussgesang (als Teil eines freimaurerischen Texts), was z. B. brüderlich erklärt): - 1) Nun auf und lasst verlauten, Ihr brüderlich Vertrauten! Wie ihre geheim verehret, Nach außen sei’s gekehret! Nicht mehr in Sälen Verhalle der Sang. - 2) Und jubelnd übermaßen Durchziehet neue Straßen! Wo wir ins Leere schauten, Erscheinen edle Bauten Und Kranz an Kränzen Die Reihen entlang, - 3) So äußeres Gebäude Verkündet inn‘re Freude; Der Schule Raum erheitert, Zu lichtem Saal erweitert; Die Kinder scheuen Nicht Moder noch Zwang. Alle sechs Strophen haben denselben prosodischen Aufbau und alle letzten Verse reimen sich, was alle Strophen durch prosodische Kohäsion verkettet. Aber jede Strophe hat ihre eigenen Reime und ist semantisch intern kohärenter als über jede Strophe hinaus. So ist der Beginn von Strophe 2) durch den Bezug von Jubel auf Sang semantisch relativ schwach verbunden. Hingegen stellt Gebäude im ersten Vers von Strophe 3) eine starke Kohärenz und Kohäsion zu Bauten her. Text- und Diskursverkettung: Ein Beitrag zur Intertextualität 191 <?page no="192"?> D) Ein Zeitungstext verkettet meist eine große Zahl verschiedener Texte. Ihre Anordnung wird durch die Wichtigkeit der Nachrichten, ferner durch die Anordnung zuerst Nachricht, dann Kommentar bestimmt, ferner durch zeitungs‐ spezifische Konventionen, wie Nationales, Internationales, Lokales, Wirtschaft, Kultur, Sport, Fernseh- und Radioprogramme, Wettervorhersagen. E) Im mündlichen Diskurs ist das Pausenkriterium nicht immer ein verlässliches Kriterium, um Textbzw. Diskursverkettung von Antworten oder Kommentaren innerhalb desselben Diskurses zu unterscheiden, wie die folgende Episode aus dem Leben des überaus schweigsamen tschechischen Komponisten Antonín Dvorak zeigt, als er einmal mit einem Freund mit der Bahn von Prag nach Wien unterwegs war. An der Grenze zwischen Böhmen und Niederösterreich beklagte sich der Freund über die Mückenplage. Erst als der Zug schon in Floridsdorf, einem damaligen Vorort von Wien war, antwortet Dvorak mit dem Kommentar „Das kommt von den Teichen“ (welche im böhmisch-niederösterreichischen Grenzbereich sehr zahlreich sind). F) Texte bzw. Diskurse können wiederverkettet werde, so in Grimms Märchen, oder vorherige oral poetry in Homers Ilias und Odyssey: Spuren der Wiederver‐ kettung zeigen sich in Homers Verteilung seiner Berichte über seine beiden menschlichen Protagonisten, Odysseus, und dessen Sohn Telemachos als den zweiten menschlichen Protagonisten: sooft einer der beiden aktiv ist, scheint der andere untätig zu sein. Eine nur inhaltliche Textverkettung stellt Vergils Aeneis dar, insofern als dieser Motive aus Odyssee und Ilias mit weiteren Motiven wiederverkettet: Viele Gesänge sind entweder nur Odysseus oder nur Telemach gewidmet; daher ist ein Telemachgesang nicht mit dem unmittelbar vorangehen‐ den Odysseusgesang wiederverkettet, sondern mit dem letzten Telemachgesang. Damit sind wir aber schon eher im Bereich der von Adamzik (2011: § 8.2) beschriebenen-Textvernetzungen als einer Textverkettung. G) Es kann zwei Ebenen der Textverkettung geben, wie bei Johann Sebastian Bachs Weihnachtsoratorium: Die erste Verkettung bewirkt die Anordnung der Rezita‐ tive, Arien und Chöre (deren Texte von verschiedenen Autor: innen stammen) jeden Tags, darüber erfolgt auf einer zweiten Ebene die Verkettung der sechs Teile des Weihnachtsoratoriums, von denen jeder einem spezifischen Tag in deren chronologischer Reihenfolge gewidmet ist. H) Die Anordnung der verketteten Texte bzw. Diskurse ist immer konventionell, entweder einer kulturellen Tradition folgend oder den Intentionen des: der jeweiligen Autor: in. Dabei finden wir eine Analogie zum Gegensatz zwischen dem unmarkierten (weil ikonischen) textinternem ordo naturalis (d. h. Korre‐ lation zwischen der berichteten Reihenfolge von Geschehnissen und ihrer fortlaufenden Darstellung im Text) und dem markierten (weil antiikonischen) ordo artificialis: Bach folgt in seinem Weihnachtsoratorium dem ordo naturalis, wie schon in G) ausgeführt. I) Im Kontrast dazu können moderne Filme durch Flashbacks einen ordo artificialis herstellen, weil die Reihenfolge der gezeigten Szenen gegen die tatsächliche Reihenfolge der Ereignisse verstößt. J) Briefwechsel und Dementis stellen hingegen eine viel losere, ebenfalls natur‐ gegebene Textverkettung her (kooperative Intertextualität nach Krause 2000), ähnlich Parodien. 192 Wolfgang Dressler <?page no="193"?> K) Scheherazade: Eine andere, sehr spezifische Art der Textverkettung finden wir in den Geschich‐ ten der Scheherazade, der Tochter des Großwesirs des persischen Königs, inner‐ halb der Rahmenhandlung der persischen Geschichten von Tausendundeiner Nacht. Der kognitive Grund der Textfolge ist der Wunsch von Scheherazade zu überleben (d. h. nicht vom Perserkönig getötet zu werden), was nur möglich ist, wenn sie in der Nacht dem König in Gegenwart ihrer Schwester Dinarzade eine spannende Geschichte erzählt, welche sie entweder am Morgen unterbricht oder nach Abschluss dem König eine weitere spannende Geschichte für die kommende Nacht verspricht. Zwischen den verketteten Texten herrschen un‐ terschiedliche Grade der Kohärenz und Kohäsion. Die Textverkettung schließt mit der Zusammenfassung: So vertrieb sie tausend‐ undeine Nacht hindurch mit ihrer Schwester dem Sultan die Zeit. Als sie dann auch die letzte ihrer Geschichten, die ähnlich wie ein Fortsetzungsroman verkettet sind, erzählt hatte, warf sich Scheherezade dem Sultan zu Füßen und sagte: „Mein Herr und Gebieter, jetzt habe ich dir alle Geschichten erzählt, die ich kenne, und ich merke dir an, dass sie dir gefallen haben. Nun bitte ich dich, schenke mir zum Lohn für dieses Vergnügen mein Leben.“ Dem König hatten die Erzählungen längst von seiner wilden Verbitterung geheilt. Er liebte dieses schöne Mädchen und glaubte wieder an das reine Herz einer Frau. Mit seinen Händen zog er Scheherezade zu sich empor und sagte: „Dich hat Allah zu mir geschickt, um mich von meinem Wahn zu befreien. Du sollst meine Frau sein und noch lange mit mir in Glück und Freuden leben.“ Darauf beschenkte der Sultan den glücklichen Großwesir, der ihm seine Tochter fast geopfert hätte. Dann schickte er eine Nachricht zu seinem Bruder Schahseman und bot ihm darin Dinarzade als Frau an, die ebenso schön und fast so klug wie ihre Schwester war. Von seinem fernen Reich Samarkand in Persien kam Sultan Schahseman so schnell wie möglich herbei. Das prächtige Hochzeitsfest der beiden Brüder mit den schönen Töchtern des Großwesirs wurde noch lange Zeit von den Dichtern besungen. Viele Tage lang jubelte das Volk, weil die Zeit des Schreckens endlich vorbei war. Die schönsten und abenteuerlichsten Geschichten, die Scheherezade dem Sultan erzählt hat, wurden für alle Zeiten und alle Menschen in den Märchen von Tausendundeiner Nacht festgehalten. 4 Zusammenfassung und Ausblick Im theoretischen Rahmen der kognitiv und semiotisch begründeten Natürli‐ chen Linguistik habe ich anhand von verschiedenen Texttypen Prinzipien der Textverkettung vorgestellt, ebenso Textwiederverkettung und bei Johann Sebastian Bachs Weihnachtsoratorium zwei Ebenen der Textverkettung kon‐ trastiert. Meist herrscht auch bei Textverkettung der ikonische ordo naturalis. Textverkettung ist konventionell oder von ihrem Autor bewusst intendiert, unterscheidet sich also klar von naturgegebenen Abfolgen der Textvernetzung wie zuerst Originaltext, dann Übersetzung oder der Reihenfolge von Briefen bei Briefwechsel zwischen denselben Personen. Mein Ausblick besteht einerseits in der Hoffnung, dass der neue Begriff der Textverkettung den terminologiewissenschaftlichen Ansprüchen des Jubilars Text- und Diskursverkettung: Ein Beitrag zur Intertextualität 193 <?page no="194"?> genügt, andererseits im Verweis auf eine im Gang befindliche Zusammenarbeit von mir und Martin Reisigl an text chaining von englischen Texten und in der Hoffnung, dass nicht nur Vertreter: innen der Textlinguistik, sondern auch der Translatologie, der Literaturwissenschaft und anderer Kulturwissenschaften meine Konzeption für anwendbar und erweiterbar halten. Bibliographie Adamzik, Kirstin (2004). Textlinguistik. Eine einführende Darstellung. Tübingen: Nie‐ meyer. Adamzik, Kirstin (2016). Textlinguistik. Grundlagen, Kontroversen, Perspektiven. Ber‐ lin/ Boston: de Gruyter. Beaugrande, Robert de/ Dressler, Wolfgang Ulrich (1981). Einführung in die Textlinguis‐ tik. Tübingen: Niemeyer. Dressler, Wolfgang Ulrich (2004). Textlinguistik mit und ohne Kontext: Diskurs versus Text. In: Oswald Panagl, Ruth Wodak (Eds.). Text und Kontext. Nürnberg: Königshau‐ sen & Neumann, 11-22. Dziubalska-Kołaczyk, Katarzyna/ Donegan, Patricia/ Dressler, Wolfgang Ulrich (Eds.) (2025). The Cambridge Handbook of Natural Linguistics. Cambridge: Cambridge University Press. Genette, Gérard (1993). Palimpseste. Frankfurt am Main: Suhrkamp. Gruber, Helmut/ Rheindorf, Markus/ Wetschanow, Karin/ Reisigl, Martin/ Muntigl, Pe‐ ter/ Czinglar, Christine (2006). Genre, Habitus und wissenschaftliches Schreiben. Eine empirische Untersuchung studentischer Texte. Münster/ Hamburg: LIT Verlag. Klein, Josef (1991). Politische Textsorten. In: Klaus Brinker (Ed.). Aspekte der Textlingu‐ istik. Hildesheim, 245-278. Kolb, Waltraud/ Mattiello, Elisa/ Dressler, Wolfgang Ulrich (2023). Human and machine translation of occasionalisms in literary texts: Johann Nestroy’s Talisman and its English translation/ s. Target: International Journal of Translation Studies. Krause, Wolf-Dieter (2000). Der interlinguale Vergleich. In: Wolf-Dieter Krause (Ed.). Textsorten. Kommunikationstypologische und konfrontative Aspekte. Frankfurt am Main: Lang, 119-143. Reisigl, Martin/ Wodak, Ruth (2001). Discourse and Discrimination. Rhetorics of Racism and Antisemitism. London: Routledge. Reisigl, Martin (2011). Kommunikationstypologien zum Handlungsbereich der Politik. In: Stefan Habscheid (Ed.). Textsorten und sprachliche Handlungsmuster: Linguistische Typologien der Kommunikation. Berlin: de Gruyter, 437-472. 194 Wolfgang Dressler <?page no="195"?> Organizational Translation Competence: Reframing Expertise Peter Sandrini Abstract: The rapid ascent of Neural Machine Translation (NMT) and Large Language Models (LLMs) has placed the translation profession at a critical juncture, fueling anxieties regarding automation and the erosion of traditional expertise. This paper proposes a paradigm shift from lang‐ uage-centric models toward "Organizational Translation Competence" (OTC). While traditional competence focuses on linguistic transfer, OTC emphasizes the individual’s capacity to architect and manage complex translation ecosystems. By integrating business management, quality assurance, terminology management, and AI literacy, OTC reframes the translator as a strategic professional capable of optimizing workflows and mitigating technological risks. Ultimately, this framework offers a pathway for human-machine collaboration, repositioning translators as essential architects of global communication in an automated era. 1 Introduction The field of translation, both as a practice and an academic discipline, stands at a critical juncture. Rapid technological advancements, particularly in auto‐ mated translation powered by neural machine translation (NMT) and large language models (LLMs), are fundamentally reshaping the industry and the very definition of translation competence. This technological surge, while offering unprecedented potential for efficiency and scale, has simultaneously fueled a pervasive anxiety regarding the future of professional translation and its viability as a sustainable career. Compounding this challenge is a perceived decline in the public regard for the translation profession, coupled with increasing pressure on practitioners to acquire new skills and adapt to novel workflows. Terminology research has always focused on the economic and commercial requirements of practice: <?page no="196"?> Die Motivationen, die zu diesen Forschungsarbeiten führen und nicht selten auch explizit angegeben werden, ergeben sich oft aus der Praxis in Wirtschaft und Handel, Recht, Justiz und Verwaltung. (Budin 2019: 8) The motivations that lead to this research, which are often explicitly stated, often arise from practice in business and commerce, law, justice and administration. (my translation) Terminology work and Terminology Management as a subarea of translators’ work are therefore less affected by these existential challenges. Traditional conceptions of translation expertise, however, have often centered heavily on linguistic proficiency and cultural knowledge, and may require a similar shift in the focus of research and training. The contemporary landscape demands a broader perspective. As automation increasingly handles routine linguistic transfer, the value proposition for human translators is shifting towards more complex, strategic, and organizational functions. For training and professional qualification, this necessitates moving beyond a narrow focus on scholarly or theoretical subjects, to embrace a more holistic understanding of the skills required to navigate and manage multilingual communication effectively in organizational contexts. This chapter introduces and elaborates upon the concept of Organizati‐ onal Translation Competence (OTC). OTC represents a comprehensive framework that transcends textproducing skills, encompassing an individual’s capacity within an organization to architect the strategic management and leveraging of translation processes and resources to achieve global objectives. It involves a synergistic integration of a set of strategic capabilities such as resource management, workflow optimization, quality assurance paradigms, and sophisticated technology integration. This research argues that developing such expertise is crucial for translation experts to contribute to organizations seeking to enhance global competitive‐ ness, improve operational efficiency, and mitigate the inherent risks of language barriers in international operations. Furthermore, OTC provides an opportunity for proactively addressing the challenges and opportunities presented by au‐ tomation, fostering productive human-machine collaboration, and cultivating specialized, highvalue translation-related expertise that complements rather than competes directly with automated systems. 196 Peter Sandrini <?page no="197"?> 2 The Evolving Language Industry: Contextualizing the Need for OTC The language industry is undergoing a profound transformation, driven pri‐ marily by technological innovation and evolving market demands. Despite anxieties about automation, according to some sources the industry continues to demonstrate robust annual growth: “This year it has again grown by about 11%” (Faes and Massey 2024: 20). Significantly, technology and related services account for a large and growing portion of this market value. Faes and Massey note a clear trend: The biggest is what we call technology […]. Typically, the techier the players are, the more successful they are as well. (Faes and Massey 2024: 25) This highlights a fundamental shift where technological proficiency is no longer merely advantageous but increasingly central to success in the language services sector. Within the core service of translation itself, a major reconfiguration is underway. The traditional Translation Memory (TM) based workflow is steadily giving way to machine translation postediting (MTPE): The language industry has had to transition from a Translation Memorybased workflow to an MTPE (machinetranslation postediting) workflow. (Faes and Massey 2024: 25) MTPE now constitutes the ‘biggest portion’ of core translation services: “By now, the biggest portion of this core service is postediting, which has become huge” (Faes and Massey 2024: 21). This transition is reshaping the daily work of translators, with predictions suggesting that by 2025, a significant majority of translators’ work will shift from creating translations to reviewing and editing machine-generated output: In their 2022 research report, Gartner reckons that by 2025 enterprises will see 75 % of the work of translators shift from creating translations to reviewing and editing machine translation output. (van der Meer 2024: 40) This evolution impacts not just translation but also adjacent services like localization, where technical aspects such as testing are becoming potentially larger components than the purely linguistic task itself: It is no longer a simple matter of ‘here is Microsoft Word in English, now localize it into German’. The vertical has become a lot more complex, and although the language Organizational Translation Competence: Reframing Expertise 197 <?page no="198"?> component per se remains a sizable portion of it, localization testing and all that comes with it is potentially even larger. (Faes and Massey 2024: 22) The actual act of linguistic transfer is increasingly seen as readily available, often near-instantaneous. Consequently, human effort and associated costs are shifting towards the preparation and oversight of the translation process: With the implementation of these practices and the right quality and volume of data, new translation costs could become nearzero and capacity infinite, although there will be a cost to obtain and clean that data. (van der Meer 2024: 37) As Pym (2025) notes, these technologies push translators … towards more active decisionmaking roles in crosscultural communication, beyond the sentencereplacement processes that are now more or less automated. (Pym 2025: 1) This evolving industrial context underscores the inadequacy of traditional, languagecentric models of expertise and necessitates a framework like OTC that incorporates strategic, managerial, and technological dimensions. 3 Integrating Translation Expertise While foundational linguistic and cultural competences remain essential, they are no longer sufficient. Different approaches in Translation Studies propose varying definitions for the two concepts of competence and expertise: “Com‐ petence and expertise have emerged as distinct yet interrelated concepts for understanding translation capacity” (Sun and Muñoz Martin 2025: 43). An expanded concept of translation expertise, centered on organizational capabilities, has the potential to bolster the professional standing of translators. This requires embracing technology not as an adversary, but as an integral component of expert knowledge. Echoing Jacques Ellul’s insights on technology, it is not technology itself that dictates outcomes, but how it is wielded; it can be liberating or enslaving depending on its application and control: … c’est que la technique est libératrice. Ceux qui l’utilisent sont les esclavagistes. L’ouvrier n’est pas victime de la technique, mais de ses maîtres. (Ellul 1955: 425). According to the understanding of competence as “a set of different capacities and skills necessary for completing a translation or interpreting task” (Tiselius and Hild 2017: 426), we can broaden this definition by substituting the last part for the skills necessary for planning and organizing translation or interpreting tasks. This allows us to conceptualize a broader expertise. For this, it is useful to distinguish between the individual and the institutional level. 198 Peter Sandrini <?page no="199"?> An earlier concept, Institutional Translation Competence (ITC) (Sand‐ rini 2019), refers to the overall translation capability of an organization: “Ever‐ ything within the boundaries of an organization that enables its performance belongs to its competence” (Taatila 2004: 12). It encompasses all supraindividual aspects: the working environment, available tools, management structures, processes, planning, quality guidelines, and technological infrastructure that shape translation activities within an institution or organization. Schnabel et al. aptly described this as the translation architecture of an organization which is “responsible for translations on behalf of a company, across departments, deli‐ verables, and formats” (Schnabel et al. 2015: 1). The proficiency and performance of individual translators within an organization, and the quality of translation services become, in part, products of these institutional structures and policies. Organizational Translation Competence (OTC), as defined here, opera‐ tes primarily at the individual level but is intrinsically linked to building insti‐ tutional capacity. OTC represents the individual’s ability to organize translation productively and successfully - the competence of translation architects who … are expected to know all aspects of the workflow, tools, schedules, roles, financial impacts, and technology stack. This person owns the success or failure of the translation strategy. (Schnabel et al. 2015: 1) While ITC relies on the OTC of its staff, it transcends the sum of individual competencies through networking, collective resource utilization, and insti‐ tutionalization (see Table 1). An organization might possess highly skilled individual translators, but without effective ITC, which encompasses elements like joint planning, shared resource management, and optimized workflows, it may struggle to achieve the consistency, efficiency, and productivity levels of an organization with strong ITC, even if the latter employs translators of only average individual linguistic skills. OTC, therefore, focuses on the individual’s capacity to design, manage, and optimize translation processes and environments (including technology and workflows), forming the bedrock upon which effective ITC is built. Organizations strive to develop robust ITC by institutionalizing these capabilities, thereby ensuring operational continuity and reducing dependency on specific key individuals. Organizational Translation Competence: Reframing Expertise 199 <?page no="200"?> ITC OTC Level Institutional / Organizatio‐ nal Individual Focus Implementation / Architec‐ ture Process Design / Capabi‐ lity Orientation Longterm / Strategic Aspirational / Develop‐ mental Nature Collective / Systemic Personal / Skillbased Goal Standardized Efficiency & Quality Ability to Design & Ma‐ nage Processes Tab. 1: Comparison of ITC and OTC This distinction highlights the need to adapt contemporary translator training curricula to cultivate OTC alongside traditional skills, thereby fostering a more comprehensive and modern understanding of translation competence required in professional settings. 4 Organizational Translation Competence The concept of Organizational Translation Competence (OTC) contrasts with common lay perceptions of the translation profession, which often reduce the field to the act of individuals translating text, and frequently involve compari‐ sons between human performance and machine output. OTC as professional translation expertise, in contrast, involves the ability to establish and manage an optimal environment for translation production, regardless of whether the core task is performed by human translators, machine translation systems, or a combination thereof. OTC extends, or at least complements, traditional translation competence models, such as those developed by PACTE (2008) and the European Master’s in Translation network (EMT 2022). In the present EMT Framework, communication and organisational skills are mentioned under Personal and Interpersonal skills: 21. Plan and manage time (including complying with deadlines) 22. Manage workload, cognitive load, stress and critical professional situations 23. Work both autonomously and in virtual, multicultural and multilingual teams, using appropriate communication technologies 24. Use social media responsibly for professional purposes 200 Peter Sandrini <?page no="201"?> 25. Take account of and adapt the organisational and physical ergonomics of the working environment 26. Continuously selfevaluate, update and develop competences and skills through personal strategies and collaborative learning and acknowledge the importance of lifelong learning (EMT 2022: 10) These established models primarily focus on the skills required to perform the act of translation - what PACTE describes as “the underlying system of knowledge needed to be able to translate” (PACTE 2008: 47). In contrast, OTC emphasizes the strategic and organizational skills required to manage and architect the entire translation ecosystem. While foundational linguistic and cultural transfer competence is presupposed, OTC specifically encompasses expertise across several key organizational domains as outlined below. 4.1 Business Management Business management related to translation involves the strategic planning, organizing, directing, and controlling of an organization’s translation-related resources (human, financial, technological) to ensure efficient and effective operations. Often overlooked in traditional translator training courses, these skills are central to OTC. Key areas include: - Strategic Planning: Analyzing market trends, assessing internal capabi‐ lities, setting longterm translation goals, and developing strategies for process optimization and technology adoption. - Leadership & Human Resources: Motivating teams, fostering positive work environments, recruiting, training, and managing personnel with the right mix of skills (not just linguistic) to enhance overall institutional competence. - Financial Management: Budgeting for translation activities, analyzing costs and ROI of different approaches (e.g., MT vs. human, different tool‐ chains), and ensuring financial stability related to language services. - Project Management: Planning, executing, monitoring, and finalizing complex translation projects within scope, time, and budget constraints. - DecisionMaking & ProblemSolving: Analyzing complex situations (e.g., choosing technology stacks, addressing quality issues), evaluating options, making sound organizational decisions, and effectively resolving problems using critical thinking. - Communication & Adaptability: Effectively communicating with di‐ verse stakeholders (clients, translators, engineers, management), building Organizational Translation Competence: Reframing Expertise 201 <?page no="202"?> relationships, resolving conflicts, and adapting strategies and processes in response to the rapidly changing technological and market environment. 4.2 Quality Management As translation production becomes increasingly automated, managing quality shifts from a purely posthoc evaluation to an integral part of the entire work‐ flow. Quality Management (QM) within OTC focuses on meeting stakeholder requirements. This involves activities defined by Lommel and Melby as the integration and coordination of management activities focused on ensuring the organization fulfills stakeholder requirements predictably, consistently, and reliably. (Lommel and Melby 2018: 4) This focus on quality spans the entire translation process, which Vela Valido (2021: 97) subdivides into the following stages: - Preproduction: Quality planning (defining standards, metrics, processes) and quality estimation. - During Production: Quality Assurance (QA - process-oriented, preven‐ ting defects) and Quality Control (QC - product-oriented, identifying defects), including managing editing and revision workflows as mandated by standards like ISO 17100 (2015: 70). This increasingly involves designing and managing MTPE workflows. - Postproduction: Quality Assessment/ Evaluation, including Machine Translation Evaluation (MTE) methodologies. Expertise in QM frameworks (e.g., MQM) and their practical implementation is vital for ensuring the reliability and fitnessforpurpose of translation outputs in diverse contexts. Editing and revision are also required by international standards for translation services (e.g., ISO 17100 2015) where they are included in the translation workflow (Mellinger 2018: 310). Thus, QM is a vital part of a translator’s organizational expertise. 4.3 Terminology Management Drawing on Wright and Budin (1997), Weilandt (2014) defines Terminology Management as Terminologiemanagement umfasst alle aufeinander abgestimmten Maßnahmen zur Planung, Steuerung, Verwaltung, Bereitstellung und Kontrolle der spezifischen ter‐ minologischen Festlegungen eines Unternehmens (Weilandt 2014: 225) 202 Peter Sandrini <?page no="203"?> Terminology Management encompasses all coordinated measures for the planning, control, administration, supply and monitoring of a company’s specific terminological definitions. (my translation) Within OTC, this involves: - Strategic Planning: Defining the role and objectives of terminology work within the organization’s broader communication and knowledge management strategy. - Infrastructure & Control: Establishing and managing the necessary infrastructure (databases, tools) and processes for consistent terminology creation, validation, and maintenance. - Supply & Integration: Ensuring terminology resources are accessible and seamlessly integrated into translator workflows, including machine translation systems and authoring environments. - Monitoring & Quality: Continuously monitoring terminology usage and verifying its accuracy and consistency across all organizational content. Ef‐ fective Terminology Management is crucial for brand consistency, technical accuracy, translation efficiency (especially with MT), and overall commu‐ nication quality. OTC involves understanding Terminology Management not just as a linguistic task, but as a critical component of organizational knowledge architecture. 4.4 Automation Proficiency in leveraging translation technologies is fundamental to OTC. This extends beyond userlevel competence with Translation Memory tools to encompass a strategic understanding and management of automation. Thus, the translator becomes a “solutions consulting and director of technology solutions, cloud solutions architect, or solutions manager for machine intelligence” (Pym 2021: 51). Key skills include: - Machine Translation: For Krüger and Hackenbuchner (2022), MT Literacy consists in technical, linguistic, economic, societal, and cognitive MT exper‐ tise. MT literacy embedded into OTC enables translators to manage data for training and finetuning, integrate MT into workflows, and evaluate MT output quality. - LLMs: Again, we relate to Krüger (2023) and the AI Literacy framework, which means developing a critical understanding of AI principles, capabili‐ ties, ethical considerations, and limitations within the context of language processing and translation. Proficiency with emerging AI tools, including Organizational Translation Competence: Reframing Expertise 203 <?page no="204"?> LLMs, involves setting up and managing local and private LLMs for security and customization, managing large datasets, tailoring foundational models, and deploying these tools effectively within the translation environment of an organization. This includes understanding the nuances of humanma‐ chine interaction in translation, recognizing that machines lack the shared world knowledge, consciousness, and intentionality inherent in human communication (Holste 2024: 211). The focus shifts to leveraging AI for efficiency while applying human oversight for quality, nuance, creativity, and critical judgment. - Core Translation Technologies: Managing and integrating traditional tools like Translation Memory systems effectively informs decisionmaking about translation tools adoption and use. 4.5 Data Management Data are crucial for terminology as well as for all forms of automation, and interoperability of data constitutes a key aspect; be it multilingual text corpora, aligned sentence pairs, or multilingual terminology resources. All this data must be collected, analyzed, evaluated, processed, and organized. Thus, linguistic competence combined with technical skills creates new roles for translators: “specialized roles - particularly Data Curators - will emerge” (Volk and Läubli 2024: 101). The prerequisite for this is wellfounded knowledge about data, highlighting comprehensive data literacy (Krüger and Hackenbuchner 2022: 396) as an integral part of OTC. 5 The Role of Humans A sustainable role of human translators cannot be defined by competing with evolving machines in the production process. As many manufacturing areas have shown, automation leads to a streamlining of production processes and important changes in job profiles. Thus, translators have to decide on which side of this change they will stand: Competing with machines on the one hand, or gaining control of the process on the other hand. Human translators will need to adapt to rapid technological advancements, integrating new AI tools and methodologies in their practices. Automation, data management and tools proficiency clearly draw upon the concept of Computational Translation Studies, described by Budin as a 204 Peter Sandrini <?page no="205"?> … juncture, where digital humanities research with its history in machine translation and literary and linguistic computing meets language industry with its use of trans‐ lation and language technologies and tools, and multilingualism. (Budin 2015: 427) This confluence of multilingualism and humanities on one side, and technology and automation on the other side means also a shift in importance related to competences and skills. Technological competence as a decisive element in OTC is growing in importance: “programming skills will continue to be more important than language skills in many roles” (Volk and Läubli 2024: 101), and languagerelated computer skills are requested by the industry increasingly more than pure language skills. From reviewing the skills and competencies […] it appears that technology-related competencies are the most prominent. Translation skills or knowledge of the proces‐ ses in the language industry are considered a plus. (Kappus 2024: 136) The declining number of students in translator training programs may also be caused by a fundamental change in expectations and requirements. Traditional mindsets of high school graduates interested in Translation Studies, focus on language, culture and interpersonal communication, which they perceive to be increasingly taken over by technology and automation, and therefore training seems pointless to them. On the other hand, new requirements by the industry give preference to technology and human-machine interaction, which in turn makes choosing Translation Studies less appealing to many of those young candidates. Adapting the training profile in line with the abovementioned orientation could lead to attracting new target groups of students. In a highly technological environment, the role of human translators comprises more than mastering “knowledge needed to be able to translate” (PACTE 2008: 47). They rather … should act as language services advisors or language consultants, advising their customers on the best approach to a particular assignment and explaining the benefits or drawbacks of certain translation methods. (FIT 2017: 2) Acting as “translation workflow engineers” (Hummel 2019: 20), languageservi‐ ces advisors and overall organizers of translation tasks professionals “must believe that technology - including machine translation - creates an opportu‐ nity, not an obstacle” (Melby and Hague 2019: 210). Organizational translation competence can provide a solid basis for this, including risk assessment compe‐ tence, consulting competence, and service competence. Organizational Translation Competence: Reframing Expertise 205 <?page no="206"?> 6 Outlook The landscape of translation is being irrevocably altered by technological advan‐ cements and shifting industry demands. Pure linguistic transfer is increasingly commodified by automation, pushing human expertise towards higherorder functions. This paper has argued for the concept of Organizational Translation Competence (OTC) as a necessary evolution in our understanding of translation expertise - a move to encompass the strategic, managerial, quality-focused, and technological skills required to architect and orchestrate translation processes effectively within organizations, companies or institutions. After several disciplinary turns in the history of Translation Studies perhaps a new turn is needed, focusing on technology and business requirements, a direction which Gambier (2014: 8) described as a trend “towards an Economic Translation Studies” by formulating different thematic spaces where economics play a decisive role in translation. Similarly, Carreira advocates an “industrial turn” in Translation Studies: This shift entails delving into the economic and business facets that impact professio‐ nal translation, drawing upon disciplines like Economics or Market Research, among others. (Carreira 2024: 198) OTC differentiates from, yet complements, Institutional Translation Compe‐ tence (ITC) by focusing on the individual’s capacity to manage the translation ecosystem. Its core components - business acumen, quality management exper‐ tise, strategic Terminology Management, and automation/ AI literacy - repre‐ sent the skill set needed to navigate the complexities of modern multilingual communication. The development of OTC is crucial not only for individual career sustainability but also for organizations seeking to leverage translation effectively for global success. Embracing OTC requires significant shifts in translator training, professional development, and industry perception. It necessitates fostering a positive and critical engagement with technology, recognizing it as a powerful tool to be managed and integrated, rather than an existential threat. Open dialogue among translation industry stakeholders and the exploration of busi‐ ness models that integrate rather than fragment the role of translators across domains are considered here to be more productive responses to advances in technology than giving in to automation anxiety. (NunesViera 2020: 16) By cultivating Organizational Translation Expertise, the translation profession can redefine its value proposition, moving from pure linguistic equivalences to 206 Peter Sandrini <?page no="207"?> strategic communication management, thereby securing a vital and evolving role in an increasingly interconnected and automated world. Bibliography Budin, Gerhard (2019). Zum Entwicklungsstand der Terminologiewissenschaft. In: Dre‐ wer, Petra/ Pulitano, Donatella (Eds.). Terminologie: Epochen - Schwerpunkte - Umsetzungen, Kommunikation und Medienmanagement. Zum 25-jährigen Bestehen des Rats für Deutschsprachige Terminologie. Berlin: Springer Vieweg, 7-19. Budin, Gerhard (2015). Digital humanities, language industry, and multilingualism: Global networking and innovation in collaborative methods. In: CIUTI-Forum 2014: Pooling Academic Excellence with Entrepreneurship for New Partnerships. Peter Lang, 423-448. Carreira, Oliver (2024). Neoliberal practices endangering the sustainability of the trans‐ lation profession: Non-standard forms of work, labor monopsonies, and technology as a catalyst for precarization. In: Amir J. Daghigh, Michael Shuttleworth (Eds.). Translation and Neoliberalism. Cham: Springer, 187-202. EMT (2022). European Master’s in Translation - EMT Competence Framework. Brussels: European Commission. Available at: https: / / commission.europa.eu/ system/ files/ 2022 -11/ emt_competence_fwk_2022_en.pdf (Accessed: 30 November 2022). Faes, Florian/ Massey, Gary (2024). Charting the language industry: Interview with an industry observer. In: Massey, Gary/ Ehrensberger-Dow, Miriam/ Angelone, Eleonora (Eds.). Handbook of the Language Industry. Berlin/ Boston: de Gruyter, 17-31. FIT (2017). FIT Position Paper on the Future for Professional Translators. Fédération In‐ ternationale des Traducteurs. Available at: https: / / wa1.fit-ift.org/ wp-content/ uploads / 2015/ 04/ Future-of-professional-translators-bilingual-exit_April-2017.pdf (Accessed: 30. November 2022) Gambier, Yves (2014). Changing landscape in translation. International Journal of Society, Culture & Language, 2(2 Special Issue on Translation, Society and Culture), 1-12. Holste, Alexander (2024). Automatisierte Wissenskommunikation. Berlin: Frank & Timme. Hummel, Jochen (2019). Sunsetting CAT. Presentation at the JIAMCATT Annual Mee‐ ting “The Shape of Things to Come“, Luxembourg, 13 May 2019. ISO 17100 (2015). Translation services — Requirements for translation services. Geneva: International Standardization Organization. Kappus, Martin (2024). Language technology developers. In: Massey, Gary/ Ehrensber‐ ger-Dow, Miriam/ Angelone, Eleonora (Eds.). Handbook of the Language Industry. Berlin/ Boston: de Gruyter, 120-141. Organizational Translation Competence: Reframing Expertise 207 <?page no="208"?> Krüger, Ralph (2023). Artificial intelligence literacy for the language industry - with particular emphasis on recent large language models such as GPT-4. Lebende Spra‐ chen, 68(2), 283-330. Krüger, Ralph/ Hackenbuchner, JoAnn (2022). Outline of a didactic framework for com‐ bined data literacy and machine translation literacy teaching. Current Trends in Translation Teaching and Learning, 375-432. Melby, Alan/ Hague, Daryl H. (2019). A singular(ity) preoccupation: Helping translation students become language services advisors on the age of machine translation. In: Sawyer, David/ Austermühl, Frank/ Enríquez Raído, Vanessa (Eds.). The Evolving Curriculum in Interpreter and Translator Education. Amsterdam/ Philadelphia: John Benjamins, 205-228. Nunes Vieira, Lucas (2020). Automation anxiety and translators. Translation Studies, 13(1), 1-21. PACTE (2008). Building a translation competence model. In: Alves, Fabio (Ed.). Trian‐ gulating Translation: Perspectives in Process Oriented Research. Amsterdam: John Benjamins, 43-66. Pym, Anthony/ Ester Torres-Simón (2021). Is automation changing the translation pro‐ fession? International Journal of Society, Culture & Language, 2021/ 270, 39-57. Pym, Anthony (2025). Risk management in translation: A reply to critics. Position paper, February 2025, Version 1.1. Sandrini, Peter (2019). Institutionelle Translationskompetenz. In: Luttermann, Claus/ Luttermann, Karin/ Kazzazi, Kerstin (Eds.). Institutionelle und individuelle Mehrsprachigkeit. Rechtslinguistik: Studien zu Text und Kommunikation, Band 10. Münster: LIT, 91-109. Schnabel, Bryan/ Hackos, JoAnn T./ Raya, Rodolfo M. (2015). A practical guide to XLIFF 2.0. Laguna Hills: XML Press. Sun, Sanjun/ Muñoz Martín, Ricardo (2025). Reframing translation expertise for the AI era. In: Sanjun Sun, Kanglong Liu, Riccardo Moratto (Eds.). Translation Studies in the Age of Artificial Intelligence. Routledge, 42-61. Taatila, Vesa (2004). The concept of organizational competence - A foundational analysis. Jyväskylä Studies in Computing, Dissertation. Tiselius, Elisabet/ Hild, Adelina (2017). Expertise and competence in translation and in‐ terpreting. In: Schwieter, James W./ Ferreira, Ana (Eds.). The Handbook of Translation and Cognition. Wiley Blackwell, 423-444. Van der Meer, Anne-Maj (2024). Evolution of the language industry. In: Massey, Gary/ Eh‐ rensberger-Dow, Miriam/ Angelone, Eleonora (Eds.). Handbook of the Language Industry. Berlin/ Boston: de Gruyter, 33-47. 208 Peter Sandrini <?page no="209"?> Volk, Martin/ Samuel Läubli (2024). MT-Developers. In: Massey, Gary/ Ehrensber‐ ger-Dow, Miriam/ Angelone, Eleonora (Eds.). Handbook of the Language Industry. Berlin/ Boston: de Gruyter, 101-119. Weilandt, Annette (2014). Terminologiemanagement. Ein prozessorientierter Ansatz am Beispiel der Automobilindustrie. Leipziger Studien zur angewandten Linguistik und Translatologie 15. Frankfurt a.M.: Peter Lang. Wright, Sue-Ellen/ Budin, Gerhard (1997) (Eds.). Handbook of Terminology Management. Volume 1: Basic Aspects. Amsterdam/ Philadelphia: John Benjamins. Organizational Translation Competence: Reframing Expertise 209 <?page no="211"?> Der Akademie-Verlag Berlin und sein translatorisches Verlagsprofil Larisa Schippel Abstract: Dieser Beitrag verortet sich am Schnittpunkt von Untersuchun‐ gen zum Wissens- und Wissenschaftstransfer, der durch Übersetzungen in Gang gesetzt und in Gang gehalten wird, und fragt nach deren innova‐ tiven oder konsekrierenden Wirkungen sowie übersetzungshistorischen Forschungen, speziell zu translatorischen Prozessen im Exil (1933-1945) und in der DDR. Damit ist die Absicht und Hoffnung verbunden, an gemeinsame Forschungsinteressen mit dem hier zu Ehrenden anknüpfen zu können, um auch weiterhin im produktiven Austausch zu bleiben. 1 Voraussetzungen Am 1. Juli 1946 wurde auf der Grundlage des Befehls Nr. 187 der Sowjetischen Militäradministration (SMAD) die Deutsche Akademie der Wissenschaften zu Berlin (DAdW) als Nachfolgerin der ehemaligen Preußischen Akademie der Wissenschaften wiedereröffnet. Im Statut vom 31. Oktober 1946 war auch vorgesehen, einen eigenen Verlag zu gründen. Die Gesamtsitzung der Akademie vom 19. Dezember beschloss am 23. Dezember 1946 die Gründung des Akade‐ mie-Verlag G.m.b.H Berlin (vgl. Links 2016: 49). Dieser Verlag gilt als der wichtigste Wissenschaftsverlag der DDR mit einer beachtlichen Bruchproduktion, die natürlich in engem Zusammenhang mit den Forschungseinrichtungen der Akademie der Wissenschaften steht, deren Forschungsergebnisse und Langzeitvorhaben in den Büchern und Fachzeit‐ schriften des Akademie-Verlags publiziert werden. Neben diesen Publikationen erscheint im Verlag aber auch eine Vielzahl von Übersetzungen, die hier an einem ausgewählten Fallbeispiel Anhaltspunkte dafür bieten sollen, von einem translatorischen Verlagsprofil zu sprechen im Sinne einer produktiven translationshistorischen Kategorie. <?page no="212"?> Diese translationshistorische Analyse basiert auf einer Objektkonstruktion im Sinne der doppelten Historisierung, wie sie die Histoire croisée in ihrer „spezifischen Verbindung von Beobachterposition, Blickwinkel und Objekt“ (Werner und Zimmermann 2002: 609) anstrebt. Im Ansatz der Histoire croisée geht es um ein Verfahren, das soziohistorische Erkenntnis von einer spezifischen Raum-Zeit-Konstellation her zu erzeugen versucht. Die zeitliche Dimension sowohl der Produktion von Erkenntnis als auch des Gegenstands der Erkenntnis ist hier in Rechnung zu stellen. Es geht also um eine doppelte Form der Historisierung, bei der der historischen Dimension des Erkenntnisprozesses der gebührende Platz eingeräumt wird, insbesondere was den Standort der Beobach‐ tung, das Beobachtungsinstrumentarium, die Perspektive und die Brennweite des Beobachtungsprozesses anlangt (Werner und Zimmermann 2002: 609). Kurz gesagt geht es um die Relation zwischen dem Zeitpunkt des Geschehens und dem Zeitpunkt der Erkenntnis. Damit ist zugleich gesagt, dass von der Ebene der Handelnden ausgegangen wird, von den Konflikten, in denen sie standen, und den Strategien, die sie zu ihrer Lösung entwickelten. Aber es gibt noch einen weiteren Aspekt, den ich kurz thematisieren möchte, bevor ich mich auf den Gegenstand einlasse, und das ist die historisch-politische Rahmensetzung, mit der sich jede nicht primär auf Demaskierung und Entlar‐ vung des „Systems“ zielende Behandlung eines DDR-Themas konfrontiert sieht: Abgesehen von jenen Rahmensetzungen, die noch in der Wendezeit aus der Regierung Kohl kamen, wie etwa die Aufforderung des damaligen Innen‐ ministers Kanter, die Gesellschaft und die Geistes- und Sozialwissenschaftler sollten sich zur „Delegitimierung der DDR“ engagieren, und so stürzte sich die Gilde der Zeit- und Gegenwartshistoriker: innen auf die Dekonstruktion der DDR. Für den deutsch-deutschen Historikerstreit kommt Martin Sabrow, einer der führenden Historiker auf dem Gebiet der Geschichtswissenschaft, der bis zur Wende als Geschichtslehrer an einem Westberliner Gymnasium gearbeitet hatte, am Beispiel der deutsch-deutschen Geschichtswissenschaft zu dem Schluss: So wichtig […] die jahrelange Debatte um Leistung und Versagen der DDR-Ge‐ schichtswissenschaft für die politische und moralische Selbstverständigung der deut‐ schen Vereinigungsgesellschaft war, so unfruchtbar war sie wissenschaftlich (Sabrow 2002: 14). Und für die Arbeit am Gegenstand DDR-Geschichte zieht er die Schlussfolge‐ rung, die ich auch für meine Arbeitsweise als brauchbar betrachte: Um die Funktionsmechanismen und Geltungskräfte einer anderen „Seinsweise der Dinge“ nicht durch Überstülpung des vertrauten eigenen Bezugssystems unkenntlich 212 Larisa Schippel <?page no="213"?> zu machen, müssen wir uns von dem Glauben an die universale Geltungskraft des westlichen Geschichtsverständnisses verabschieden, so selbstverständlich es uns in der institutionalen Ordnung und in den (aus Binnensicht) gleichsam ‚natürlich‘ gebildeten Wertmustern der eigenen Vergangenheitsaneignung auch erscheinen mag und so sehr wir an seinem legitimen Geltungsanspruch festhalten mögen. Vielmehr gilt es, das kulturelle Gedächtnis einer diktatorisch verfassten Gesellschaft so zu untersuchen, dass die hinter ihr stehende symbolische Sinnwelt erkennbar wird (Sabrow 2000: 14). Damit begründet er das, was er zuvor als die zum Tragen kommende Arbeits‐ weise skizzierte, nämlich, sich dem Gegenstand mit jener ethnografischen Behutsamkeit [zu nähern], die dem Betrachter selbstver‐ ständliche Denk- und Sehgewohnheiten nicht unreflektiert dem Betrachtungsgegen‐ stand unterlegen will und die Möglichkeit nicht aus den Augen verliert, dass eine andere Vergangenheitskultur auch andersartig sein könnte (Sabrow 2000: 11). In diesem Sinne ließen sich dann die Prozesse und handelnden Personen bzw. Institutionen zum Zeitpunkt ihres Handelns in Relation setzen mit den (heutigen) Positionen des Betrachters. 2 Einbettung (Ethnografie) Ethnografisches Arbeiten - also gegenstandsbezogenes „Jagen und Sammeln“ - ist zum größten Teil auf Archive und Bibliotheken angewiesen. Diese Archive sind - das wurde bereits öfter thematisiert - nicht besonders translationsaffin, vor allem aber unvollständig. Das Archiv des Akademie-Verlags zum Beispiel verzeichnet im Register schätzungsweise bei einem Drittel der Dokumente, sie seien dem Archiv „nicht übergeben“ worden. Wer hat dort auf- und ausgeräumt? Wo liegen Interessen, Archivdokumente nicht zu übergeben? Zum Schicksal des Akademie-Verlags seit 1990 findet sich bei Chistoph Links eine Darstellung, die hier nur stichpunktartig angeführt wird: 1991 Verkauf des Verlags an die Gruppe VHC (ehemals BRD-Verlag Chemie) in Weinheim, Reduzierung der Mitarbeiter: innen von 170 auf 40; Verlagsarchiv wird an Berlin-Brandenburgische Akademie der Wissenschaften übergeben, gelangt danach ins Landesarchiv Berlin; 1996 wird die VCH-Gruppe Weinheim von John Wiley & Sons, New York, übernommen, hier nur das naturwissen‐ schaftliche Programm, das geisteswissenschaftliche Programm geht 1997 an die Oldenbourg Verlagsgruppe München, verbunden mit einem weiteren Mit‐ arbeiterabbau auf nunmehr acht Personen; 2004 wird die Oldenbourg-Gruppe von der Cornelsen Holding übernommen, Akademie-Verlag besteht als Imprint Der Akademie-Verlag Berlin und sein translatorisches Verlagsprofil 213 <?page no="214"?> 1 Dazu gehört zum Beispiel die seit 1954 vom Akademie-Verlag publizierte neue und vollständig neu übersetzte Aristoteles-Gesamtausgabe in 20 Bänden (Schippel 2023). weiter. Hier endet die Darstellung von Christoph Links. Mittlerweile werden Publikationen aus dem Akademie-Verlag in digitalisierter/ digitaler Form von De Gruyter-Brill vertrieben, da dieser Verlag das Bucharchiv des Akademie-Verlags übernahm, eine Zeitlang hielt De Gruyter das Imprint des Akademie-Verlags noch sichtbar, mittlerweile nicht mehr. 1 2.1 Stellung des Akademie-Verlags (AV) in der Verlagslandschaft der DDR Die prominente Stellung des Akademie-Verlags speiste sich natürlich aus der Autorität der Akademie und der hohen Exportleistung des Verlags, was nicht nur als Deviseneinnahme zu bewerten ist, sondern vor allem einen Reputati‐ onsgewinn für die DDR im deutsch-deutschen Verhältnis bedeutet. Naturwis‐ senschaftliche Zeitschriften wie auch die Großprojekte der Akademie trugen dazu erheblich bei. Diese Stellung des AV wurde ab Anfang der 1960er Jahre gestärkt, als die DDR an ihre Neuprofilierung der Verlagslandschaft ging. In der SBZ waren ursprünglich etwa 200 Verlagslizenzen erteilt worden, die die DDR 1951 einer Neu-Lizenzierung unterzogen hatte. Diese Neuprofilierung bedeutete die Abschaffung der Konkurrenz im Sach- und Fachbuchbereich, es wurden be‐ wusst Monopolstrukturen finanziert. (S. o. Planwirtschaft, Konkurrenz bedeutet in dieser Interpretation Verschwendung, Doppelstrukturen etc.) Nach dieser Phase gibt es rund 80 Verlage. Bei den Literaturverlagen ist die Lage etwas anders, denkt man an den Aufbau-Verlag und an Volk und Welt, aber auch die parteieigenen Verlage der anderen Parteien - Der Morgen, Union-Verlag, Verlag der Nationen, teilweise auch bei den Kirchenverlagen. Diese Konzentration und Straffung der Verlagslandschaft ermöglichte es den Verlagen, rentabel zu arbeiten, Auflagen von üblicherweise 20.000 Exemplaren zu erreichen und ohne Subventionen auszukommen - was sie wiederum vom Akademie-Verlag unterscheidet, der jährliche Subventionen erhielt - Wissenschaft ist eben nicht rentabel - auf dem Markt, für die Gesellschaft schon! Diese Situation erklärt auch das günstige Verhältnis zwischen der Zahl der Lektor: innen und der Buchproduktion: Während international ein rentabler Verlag mit einem: einer Mitarbeiter: in pro fünf Titel agiert, durchschnittlich sogar sieben Titel von einem: einer Lektor: in betreut werden, waren das bei Aufbau zwei Titel pro Mitarbeiter: in und bei Volk und Welt gar einer pro Mitarbeiter: in (Aussagen von 214 Larisa Schippel <?page no="215"?> 2 vgl. Lokatis jüngst 2023. Christoph Links, mündlich vorgetragen am 22. November 2024 auf einer Tagung am Institut für Slawistik der Humboldt-Universität zu Berlin.). Die Histoire croisée gibt bekanntlich die Instrumente des Vergleichs und des Transfers nicht auf, sondern integriert sie in ihren Bezugsrahmen. Ich werde hier jedoch keine Vergleiche vornehmen, obwohl das durchaus eine lohnenswertes Unterfangen wäre und weiter verfolgt werden sollte, um etwa zu bestimmten Fragen gezielt Vergleiche zwischen Verlagen zu unternehmen und dabei zugleich den potentiellen Status der translationshistorischen Kategorie eines translatorischen Profils von Institutionen zu bestimmen. Um tatsächlich trans-national arbeiten zu können, lag es nahe, auf die Bestände des DDR-Büros für Urheberrechte zurückzugreifen. Dort müssten die Unterlagen zu Lizenzen, Verlags- und Übersetzungslizenzen zu finden sein. Die jahrelange Korrespondenz mit dem Bundesarchiv auf Grund der Tatsache, dass diese Bestände seit Jahren als „nicht bearbeitet“ ausgewiesen werden, endete vorläufig mit der Auskunft vom 27.3.2024, dass die Bestände bearbeitet, aber noch nicht öffentlich zugänglich seien. Neben der Tatsache, dass Archive a) nicht translationsaffin sind, ist translati‐ onshistorisches Arbeiten b) abhängig von Institutionen und deren Arbeitsweise und -bedingungen. Und c) stellt sich das Problem, welchen Status haben die Fallstudien, oder anders gesagt, wie repräsentativ ist das, was sich - unter anderem auf Grund der Archivlage - beschreiben lässt. Auch hier ein Beispiel für die gegenwärtige Erkenntnissituation. In der nicht allzu umfassenden Literatur zum Übersetzen in der DDR hat sich eingebürgert, vor allem ausgehend von der Literatur- und Buchwissenschaft, einen Fokus auf die Druckgenehmigungs‐ verfahren zu legen. Diese gehörten auch zu den bereits früh freigegebenen Archivdokumenten. Sie werden schlichtweg als Zensurverfahren interpretiert, prominent durch Siegfried Lokatis 2 und den von ihm hinzugezogenen ehemali‐ gen DDR-Autor: innen und -Verlagsmitarbeiter: innen. Eine im Sinne Sabrows agierende „behutsame ethnografische“ Arbeitsweise hätte zunächst nach der Funktion dieser Verfahren in jener „anderen Sinnwelt“ zu fragen. Doch auch hier stand die Wertung vor der Analyse. Wer nach Zensur fahndet, wird sie finden. Doch primär waren diese Druckgenehmigungsverfahren umfassende Planungsverfahren. Die DDR hatte eine Planwirtschaft, und geplant wurde alles! Und wenn es einmal keine Zahnbürsten im ganzen Land gab, dann war das eben ein Planungsfehler, und der wurde im ZK der SED ausgewertet. Beim Druckgenehmigungsverfahren handelte es sich um ein ganzheitliches Verfahren, das sich als Teil der Planwirtschaft der DDR verstand. Für die Der Akademie-Verlag Berlin und sein translatorisches Verlagsprofil 215 <?page no="216"?> konkreten Vorgänge der Übersetzung von Literatur war, beginnend bei po‐ litischen Entscheidungsträgern und Institutionen sowie auf der Grundlage zwischenstaatlicher Verträge (die auch bilaterale Titellisten von zum Übersetzen vorgeschlagenen Büchern enthielten) über Fachleute, Verlegende, Literaturwis‐ senschaftler: innen, Übersetzer: innen, Lektor: innen - bis hin zu Papierkontin‐ genten, verlegerischen Entscheidungen über Auflagenhöhen, Titelgebungen, Vor- und Nachworten, zu Lizenzfragen und der deutsch-deutschen Verlags‐ kooperation, ökonomischen Bedingungen etc., ein Netzwerk von Akteuren eingebunden. Darin finden sich Hinweise auf die Aushandlungsverfahren zwi‐ schen den Beteiligten. Die Kämpfe von Ralph Schröder etwa im Verlag Volk und Welt um die Übersetzung interessanter russischsprachiger Autor: innen erklären sich genau aus diesen Aushandlungsprozessen. Und die Ergebnisse sind unterschiedlich, einmal gelingt es, ein anderes Mal nicht oder erst später, wenn sich der Wind gedreht oder der Minister gewechselt hat. Anfänglich habe ich für den Akademie-Verlag nach diesen Druckgenehmi‐ gungsverfahren im Archiv gesucht, bis ich an anderer Stelle herausfand, dass es sie für den Akademie-Verlag nicht gab. Er hatte die Akademie als Autorität im Rücken, die für die wissenschaftliche Qualität der Bücher zu bürgen hatte, er bekam ein Papier- und Druckkontingent und entschied seit 1949 autonom über die eigene Buchproduktion. Politische Eingriffe gab es wohl, wenn irgendwer Nicht-adäquates entdeckte. Wie genau sie aussahen, ist aus der Beziehung zwischen Akademie und Verlag zu ermitteln, ist aber bislang m. W. noch nicht unternommen worden. Seit 1956 verfügte die Akademie […] über drei eigene Druckereien in Altenburg (Maxim Gorki), Langensalza (Thomas Müntzer) und Gräfenhainichen (Gottfried Wilhelm Leibniz), seit 1972 mit dem Druckhaus Köthen und Kongreß- und Werbedruck Oberlungwitz sogar über fünf. Das resultierende Kombinat zählte über 1.500 Mitarbeiter, davon 220 im Verlag (Lokatis 2023: 366). 3 Zugänge für Fallbeispiele 3.1 transferorientiert In einem transferorientierten Vorgehen stellt sich immer zuerst die Frage nach der Selektion der zu übersetzenden Titel, also wer selegiert? Hier soll das zunächst am ausgewählten Titel erfolgen: Wahrscheinlichkeitstheorie: Eine Einführung / A. A. Borovkov. Berlin: Akademieverlag, 1976. Vertrieb für alle nichtsozialistischen Länder: Birkhäuser Verlag Basel. 216 Larisa Schippel <?page no="217"?> 3 Курс теории вероятностей. - М.: « Наука ». Главное издательство ФизикоФизико-математической литературы , 1972. - 288-с. - 10.000 экз . 4 Die DDR hatte vom sowjetischen Hochschulsystem übernommen, auch in der Hoch‐ schullehre mit Lehrbüchern zu arbeiten. Orig.: Курс теории вероятностей. - М.: « Наука ». Главное издательство Физико-математической литературы , 1972. - 288-S. - 10.000 Exempl. 3 An diesem Beispiel kann deutlich werden, dass man sich dieser Frage nach der Selektion am ehesten über die Person des Herausgebers nähert. Wenn man dazu weiß, dass während der Existenz der DDR wohl ca. 20.000 Studierende und Promovierende (Aspirantur) in der UdSSR studiert haben, lässt sich erklären, wo die Kontakte ihren Ursprung hatten. Hinzukommen aber, und ist vielleicht noch wichtiger, die Kooperationen zwischen den Akademien der RGW-Staaten. Die regelmäßigen Konsultationen und staatlichen Abkommen über Zusammenarbeit - natürlich in Form von Plänen, hier Fünfjahrplänen - enthalten häufig auch vereinbarte Titellisten zum Übersetzen. Eine Aufar‐ beitung dieser Titellisten und ein Vergleich zu den tatsächlich erfolgten und publizierten Übersetzungen wäre hier durchaus von Interesse. Das Original erschien 1972 in Moskau im führenden Wissenschaftsverlag Nauka. Dass der Akademie-Verlag hier ein Lehrbuch veröffentlicht, kann wohl mit der Qualität des Buches erklärt werden, sein Autor Borovkov gilt als führend auf dem Gebiet der Stochastik, und damit wiederum liegt die fachliche Begegnung mit Peter Franken nahe. Translatorisch relevant ist mitunter die Distanz zwischen Erscheinungszeit des Originals und der der Übersetzung; hier ist sie relativ klein zwischen 1972 und 1976. Da Franken Professor an der Humboldt-Universität Berlin ist, darf für die Publikation eines Lehrbuchs 4 im Akademie-Verlag auch ein studentisches Interesse an einem Einführungskurs unterstellt werden. Wir können im vorliegenden Fall also biographisch-persönliche Kontakte sowie institutionelle Kooperation als Grundlage für die Selektion annehmen. Über die konkrete Transformation, also den eigentlichen Übersetzungsprozess, will ich hier nicht sprechen. Aber Tatsache ist auch, dass auf diesem Weg sowjetische Wissenschaft über die „verbündeten“ Wissenschaftler in der DDR nicht nur in den deutschsprachigen Raum Eingang fand, sondern darüber hinaus in die internationale Wissen(schaft)szirkulation, was für sowjetische Wissenschaftler umso wichtiger war, als sie ja während des Kalten Krieges nur noch auf Russisch publizieren durften. Zu geradezu spannenden Fällen interna‐ tional verflochtener Translationsgeschichten zählt das Beispiel des ursprünglich Berliner Mathematikers Kurt Hirsch, der sich nach seinem britischen Exil in Der Akademie-Verlag Berlin und sein translatorisches Verlagsprofil 217 <?page no="218"?> 5 Exil-Trans: Leben und Arbeit verfolgter Übersetzer und Übersetzerinnen. https: / / exiltr ans.univie.ac.at/ Großbritannien seiner Russischkenntnisse entsinnt und aus dem Russischen ins Englische übersetzt und sich teilweise auf die in DDR-Verlagen erschienenen deutschen Übersetzungen stützt (Schippel 2024). Für die Frage nach einem translatorischen Verlagsprofil lässt sich als ein Bestimmungsstück gewinnen: Die Herausgeber und Übersetzer stammen überwiegend aus der jeweiligen Fach-Community. Nur noch ein Wort zur Rezeption. In einer durchaus üblichen Konstellation übernimmt ein bundesdeutscher oder in diesem Fall der Schweizer Verlag Birkhäuser aus Basel die übersetzte Fassung des Buchs von Borovkov in der Übersetzung von Rödel & Kühne und damit den Vertrieb im nichtsozialistischen Ausland. Eine Stichprobe zeigt, dass das Buch dann etwa in der British Library, der Königl. Bibliothek Kopenhagen und vielen anderen ankommt. Leider ist WorldCat für so eine Erhebung keine zuverlässige Quelle, sondern es muss immer geprüft werden, ob es sich tatsächlich um das Buch von 1976 aus dem Akademie-Verlag handelt, oder womöglich um die mittlerweile von De Gruyter vertriebene digitale Version, da De Gruyter eben die gesamte Produktion des Akademie-Verlags digitalisiert und in sein Archiv übernommen hat. Dafür hat er mittlerweile auch die Angabe des AV unter seinen Imprints gelöscht. Doch für den Lebenslauf der Übersetzung bedeutet das - wir haben das Konzept des Lebenslaufs einer Übersetzung im Exil-Transprojekt, 5 entwickelt und Stefanie Kremmel hat es dann in ihrer Dissertation zur Übersetzung des Kommunisti‐ schen Manifests angewandt -, dass diese Übersetzung wie viele andere auch seit 1976 unverändert vertrieben wird und greifbar ist. 2026 könnte sie ihren 50. Geburtstag feiern. Nebenbei scheint mir, dass die digitalisierte Form der Übersetzung diese quasi einfriert. Wären es noch Wissenschaftsverlage „alten Schlags“, wäre womöglich, wie bei schöner Literatur auch, eine Neuübersetzung mit Aktualisierung in Frage gekommen. Aber die scheinbar so mobile digitale Welt fixiert den Text gewissermaßen. 4.2 translationshistoriografisch Was bedeutet das nun translationshistoriografisch? Wenn wir Translationsge‐ schichte(n) erzählen, dann werden sie von Leser: innen deshalb „geglaubt, weil die Darstellungsmodi der Historiker[: innen] - mithin auch der Translationshis‐ toriker[: innen] - den Sinnerwartungen der Leser[: innen] entsprechen“ (Richter 2023: 19). 218 Larisa Schippel <?page no="219"?> Einer der Hauptschwerpunkte des Akademie-Verlags liegt nun eindeutig in dem, was heute als MINT-Fächer bezeichnet wird, Mathematik und Natur‐ wissenschaften, weniger Technikwissenschaften, die sind eher im Fachbuch‐ verlag zu finden. Dort wäre zu prüfen, ob sie mit Fachschulen, Technischen Universitäten, der Bergakademie etc. zusammenarbeiten. Hier also aufbauend auf der Grundlagenforschung, die vor allem an den Akademien, aber auch an Universitäten betrieben wird, daraus ergeben sich die Kontakte zu den Kollegen im Ausland, die Kenntnis von deren Forschungen und das Bemühen, sie in deutscher Sprache zugänglich zu machen, damit sie auf Deutsch dann wiederum einfließen in die internationale Wissenszirkulation. Dabei liegt der Schwerpunkt eindeutig auf Russisch als Ausgangssprache, andere Sprachen, auch Englisch, kommen vor, aber in deutlich geringerem Umfang. Interessant wäre hier der Vergleich mit anderen Wissenschaftsverlagen wie dem Verlag der Wissenschaften oder anderen Fachbuchverlagen. Daneben gibt es einen philosophisch-geisteswissenschaftlichen Schwer‐ punkt, mit einem Fokus auf den philosophischen Problemen der Naturwissen‐ schaften, sowie einen gesellschaftswissenschaftlichen Bereich. Altertumswis‐ senschaften gehören klassisch zum Kanon der Akademiedisziplinen, an der Akademie der Wissenschaften sind sie seit Kriegsende prominent besetzt, mit ihrer eigenen Übersetzungskultur (vgl. Schippel 2023). Aber es erscheinen beim Akademie-Verlag etwa auch die Volksmärchen verschiedener Länder, etwa Ungarische, Tuwinische Volksmärchen, Estnische, Lettische, Belorussische, Ukrainische Volksmärchen usw. Man sieht das Muster dahinter! Wissenschaftsgeschichte gehört naturgemäß zum Publikationsspekt‐ rum des AV. In vielen Fällen erkennt man das Interesse der Akademie-Institute, die Zusammenarbeit mit Partner: innen, die Zugehörigkeit der Herausgeber: in‐ nen und Übersetzer: innen zum akademischen Raum im weiteren Sinne. 4.3 biographisch An der Person von Peter Franken lässt sich demonstrieren, wie wissenschaft‐ liches Profil mit der Herausgeberschaft von Büchern und dem Übersetzen verflochten ist. Dazu einige kurze biografische Daten: Peter Franken wurde am 29. Mai 1937 in Moskau als Sohn deutscher Emi‐ granten geboren, 1955 übersiedelte er gemeinsam mit seiner Mutter in die DDR. 1960 schloss er sein Mathematikstudium an der Humboldt-Universität zu Berlin erfolgreich ab und absolvierte im Jahr darauf ein Zusatzstudium in Moskau. 1963 wurde er an der Universität Jena promoviert, wo er sich 1970 auch habilitierte. 1970 wurde er auch gleich zum außerordentlichen und 1972 Der Akademie-Verlag Berlin und sein translatorisches Verlagsprofil 219 <?page no="220"?> 6 https: / / mathscinet.ams.org/ mathscinet/ MRAuthorID/ 68815 dann zum ordentlichen Professor an die Humboldt-Universität berufen. Peter Franken verstarb im 53. Lebensjahr in Berlin. Den o. g. Zusammenhang will ich nur an einer kleinen Auswahl seiner publi‐ zierten Bücher belegen, weitere Angaben zum Gesamtschaffen finden sich wie immer in den mathematischen Netzwerken, beispielsweise unter MathSciNet 6 . - Als Herausgeber: • Statistische Theorie der Zuverlässigkeit: wahrscheinlichkeitstheoret. Mo‐ delle / Richard E. Barlow; Frank Proschan. In dt. Sprache hrsg. von Peter Franken. [Dt. Übers.: Eckard Knauer u. a.]. Es erscheint in der Reihe: Mathematische Lehrbücher und Monographien/ Abteilung 2 = Mathematische Monographien, Bd. 49, Teil 1, 1. Auflage 1978. Orig.: Statistical theory of reliability and life testing. • Mathematische Methoden der Zuverlässigkeitstheorie / B. W. Gnedenko; J. K. Beljajew; A. D. Solowjew. [Übers. aus d. Russ.: Eckhard Knauer.] In dt. Sprache bearb. u. hrsg. von Peter Franken. Berlin: Akademieverlag, Bde 1 + 2, 1968. Orig.: Matematičeskie metody v teorii nadežnosti. • Sowie die bereits erwähnte Wahrscheinlichkeitstheorie: Eine Einfüh‐ rung / A. A. Borowkow. Berlin: Akademieverlag, 1976. Vertrieb für alle nichtsozialistischen Länder: Birkhäuser Verlag Basel. Deutsche Übersetzung: Dr. Egmar Rödel, Dipl.-Math. Hans Kühne. - Als Übersetzer: • Aus der Reihe Arbeiten zur Informationstheorie / Von A. J. Chintschin (in 5 Bänden): Teil: 3., Ε-Entropie und Ε-Kapazität von Mengen in Funktionalräumen / Von A. N. Kolmogoroff u. W. M. Tichomirow. [Übers. aus d. Russ.: Peter Franken u. Kurt Nawrotzki] Seine eigenen Arbeiten als Autor und Mathematiker lassen sich dem o. a. Netzwerk entnehmen. 5 Resumee und Ausblick Das Interesse am Verlag und dem Autor, Herausgeber und Übersetzer Peter Franken, die ja keine Einzelfälle sind, zielte auf die Frage, ob sich nun eine übersetzungshistorische Kategorie eines translatorischen Verlagsprofils und gegebenenfalls auch anderer Institutionen gewinnen lässt, die womöglich pro‐ duktiv sein könnte? Welche Bestimmungsstücke sollte sie haben? 220 Larisa Schippel <?page no="221"?> 1. Einbettung in einen verlagsübergreifenden Kontext, allgemeine und spezi‐ fische soziohistorische Bedingungen; 2. Motivstruktur der Übersetzungen: Auftraggeber: innen, Nutznießer: innen, primär inhaltliches, hier: wissenschaftliches Motiv, andere Motive sind denkbar; 3. Gewinnung und Status der Übersetzer: innen; 4. Lebensläufe der Übersetzungen (Auflagenhöhe, Neubearbeitungen, Neu‐ übersetzungen); 5. Rezeptionsorientierungen für die Übersetzungen - innen vs. außen, hier: akademischer Raum, Exportorientierung u. ä.; 6. Faktoren des Übersetzungsprozesses - Verhältnis Lektorat/ Herausgeber‐ schaft/ Übersetzer: innen; 7. Rolle der ökonomischen Faktoren, der Rentabilität, Subventionsnotwen‐ digkeiten, wenn ja, wer subventioniert? , dazu auch Preisgestaltung und Preisentwicklung; 8. Renommee des Verlags bzw. der Übersetzungen (etwa Volk-und-Welt-Übersetzungen). Die Überlegungen, eine Kategorie des translatorischen Verlagsprofils zu entwi‐ ckeln, entstand im Vergleich verschiedener Verlage und ihres Umgangs mit Übersetzungen. Dabei zeigte sich, dass es Verlage gibt, für die Übersetzungen eine wichtige Rolle spielen oder zu bestimmten Zeiten ihrer Verlagsexistenz eine wichtige Rolle spielten. Als Beispiele aus unterschiedlichen Zeiten an unterschiedlichen Orten seien genannt: Der Piper-Verlag München, der seine Existenz u. a. mit der Übersetzung der Werke von Dostojewski (direkt aus dem Russischen) beginnt - die berühmten roten Bände; der Verlag Polirom Iași (Rumänien), der nach seiner Gründung im Jahr 1995 sein Verlagsprofil stark auf Übersetzungen stützt (vgl. Jeanrenaud (dt.) 2014) oder etwa der Wiener Zsol‐ nay-Verlag, der sogar eine eigenständige kriteriengeleitete Hierarchie seiner Übersetzer: innen begründet (vgl. Haiden demnächst). Andererseits publizieren natürlich auch viele andere Verlage Übersetzungen, ohne dass das jedoch profilbildend wirkt. Unter translationshistorischem Aspekt scheint es mir daher sinnvoll zu prüfen, ob sich eine derartige übersetzungshistorische Kategorie etabieren ließe. Der Akademie-Verlag Berlin und sein translatorisches Verlagsprofil 221 <?page no="222"?> Bibliographie Primärliteratur Statistische Theorie der Zuverlässigkeit: wahrscheinlichkeitstheoret. Modelle / Richard E. Barlow; Frank Proschan. In dt. Sprache hrsg. von Peter Franken. [Dt. Übers.: Eckard Knauer …] Es erscheint in der Reihe: , Teil 1, 1. Auflage 1978. Orig.: Statistical theory of reliability and life testing Mathematische Methoden der Zuverlässigkeitstheorie / B. W. Gnedenko; J. K. Beljajew; A. D. Solowjew. [Übers. aus d. Russ.: Eckhard Knauer.] In dt. Sprache bearb. u. hrsg. von Peter Franken. Berlin: Akademieverlag, Bde 1 + 2, 1968. Orig.: Matematičeskie metody v teorii nadežnosti Wahrscheinlichkeitstheorie: Eine Einführung / A. A. Borowkow. Berlin: Akademiever‐ lag, 1976. Vertrieb für alle nichtsozialistischen Länder: Birkhäuser Verlag Basel. Deutsche Übersetzung: Dr. Egmar Rödel, Dipl.-Math. Hans Kühne. Literatur Geschichte des deutschen Buchhandels im 19. und 20. Jahrhundert. Deutsche Demokra‐ tische Republik, Band 5, Teil 2: Verlage 2. Im Auftrag der Historischen Kommission herausgegeben von Christoph Links, Siegfried Lokatis und Klaus G. Saur, in Zusam‐ menarbeit mit Carsten Wurm (2023). Berlin/ Boston: Walter de Gruyter GmbH. Haiden, Tatsiana (2024). Dissertationsschrift: The Paul Zsolnay publishing company, 1924-1938. The translation culture. Veröffentlichung folgt 2026. Jeanrenaud, Magda (2014). Universalien des Übersetzens. Berlin: Frank & Timme. Links, Christoph (2016). Das Schicksal der DDR-Verlage. Die Privatisierung und ihre Konsequenzen. Berlin: edition berolina. Lokatis, Siegfried (1996a). Wissenschaftler und Verleger in der DDR: Das Beispiel des Akademie-Verlages. Geschichte und Gesellschaft, 22(1), 46-61. Lokatis, Siegfried (1996b). Die Gründung des Akademie-Verlags. In: Sitzungsberichte der Leibniz-Sozietät der Wissenschaften zu Berlin e. V., Band 15, 7/ 8, 81-98. Lokatis, Siegfried (2019). Zensurwerkstätten der DDR. Leipziger Arbeiten zur Verlagsge‐ schichte, Nr. 2. Leipzig. Richter, Julia (2023, zuvor Wien 2020). Translationshistoriographie. Berlin: Frank & Timme Verlag. Sabrow, Martin (2000) (Hrsg.). Geschichte als Herrschaftsdiskurs: Der Umgang mit der Vergangenheit in der DDR. Köln: Böhlau. Schippel, Larisa (2024). Exil ist Translation. Translation als „Grenzüberschreibung“. In: Kremmel, Stefanie/ Richter, Julia/ Schippel, Larisa (Hrsg.). Translation und Exil (1933-1945), III. Motive, Funktionen und Wirkungen. Berlin: Frank & Timme Verlag, 127-152. 222 Larisa Schippel <?page no="223"?> Schippel, Larisa (2025). Der ganze Aristoteles - Wissenschaft und Übersetzen. In: Chronotopos: A Journal of Translation History, 5(2), 16-31. Werner, Michael/ Zimmermann, Bénédicte (2002). Vergleich, Transfer, Verflechtung. Der Ansatz der Histoire croisée und die Herausforderung des Transnationalen. In: Geschichte und Gesellschaft, 28(4), 607-636. Göttingen: Vandenhoeck & Ruprecht GmbH & Co. KG. Der Akademie-Verlag Berlin und sein translatorisches Verlagsprofil 223 <?page no="225"?> The automatic translation of literature and literary post-editing: German versions of a short story by Kate Chopin Waltraud Kolb Abstract: Until recently, machine translation has been believed to be unable to master literary translation, but with the advent of NMT and generative AI this is no longer the case. However, there is still wide agreement that the Automatic Translation (AT) of literature, if used at all, needs to be followed by human post-editing. This contribution discusses some findings from an empirical study conducted in the spring of 2024 at the Center for Translation Studies (University of Vienna) with six students as participants. They post-edited excerpts from German translations generated by DeepL and ChatGPT of two short stories by 19 th -century American writer Kate Chopin. This contribution focuses on the AT drafts and post-edited versions of one story, “The Kiss”; it compares the performance of DeepL and ChatGPT and explores the edits made, differentiating between essential and preferential edits. Findings do not point to significant differences between the two AT systems but great intersubject variation. 1 Introduction In 2019, Gerhard Budin contributed a chapter on the 70-year long history of Machine Translation (MT) to an edited volume entitled Veränderungen in einer Generation, which discussed changes that occurred within just one generation in science, the arts, philosophy, literature, and business management. The wide-ranging adoption of MT in the translation industry over the past two or three decades certainly qualifies as such a momentous change, its impact reaching far into our everyday lives. In his contribution, Budin gave a broad overview of the status quo of MT use in the professional field, pointing out, however, that “[i]n many areas of translation, especially in literary translation, it <?page no="226"?> 1 https: / / www.statista.com/ forecasts/ 1294207/ ebook-market-revenue-worldwide 2 https: / / www.nuanxed.com/ ; https: / / reedsy.com/ will be a long time before we can speak of a meaningful use of machine trans‐ lation” (Budin 2019: 54, my translation). With recent technological advances, the situation is rapidly changing and the domain of literary translation is no longer immune to these developments. We can assume that Automatic Translation (AT) - I will use the term here to include both Neural MT (NMT) and translation via generative AI - will play an increasing role in some sectors of the book industry. The global ebook market is expected to continuously grow between 2023 and 2028 by a total of 1.3 billion US dollars (+ 9.52%) 1 , with a growing number of platforms offering ebook translations and whole new book-publishing ecosys‐ tems springing up (e.g., Nuanxed or Reedsy 2 ). At the same time, traditional publishers in many countries are also beginning to consider AT as a viable option for certain literary genres (MacDonald 2024). However, notwithstanding technological headways, there is wide agreement that automatically generated translations without subsequent human Post-Editing (PE) are still “unsuitable for translating literature” (OMC 2022, 60), which puts PE at the very heart of any investigation of the challenges and opportunities engendered by digitalization in the literary translation domain. Literary post-editing is also at the heart of this contribution. It will discuss some findings from an empirical study conducted in 2024 at the Center for Translation Studies (University of Vienna, Austria) with six MA-level students as participants. The participants post-edited German translations generated by DeepL and ChatGPT of two short stories by the 19 th -century American writer Kate Chopin. Given the limited space available, this contribution will focus on the AT drafts and post-edited final versions of one of the two stories. How does the performance of DeepL compare to that of ChatGPT? How much of the two AT drafts did the participants edit, and which kinds of edit did they make? 2 Related research The automatic translation of literature started to attract the attention of computational linguists in the early and mid-2010s. One of the primary areas of research has since been the evaluation of output generated by various types of AT systems, i.e., generic general-domain systems such as Google Translate or DeepL (e.g., Genzel et al. 2010; Webster et al. 2020) as well as MT systems that have been specially trained for literary texts (Toral and Way 2018; Kuzman et al. 2019) or texts by a particular author and/ or translator (Hansen et al. 2022; 226 Waltraud Kolb <?page no="227"?> Yirmibeşoğlu et al. 2023). More recently, the performance of generative AI for translation tasks, including literary translation, has been added to this list. In the non-literary domain, Large Language Models (LLMs) like ChatGPT have been found to achieve competitive results compared to NMT systems, especially for high-resource European languages (Hendy et al. 2023; Jiao et al. 2023; Manakhimova et al. 2023). It has also been shown that providing information on the translation task and domain in prompts tends to increase output quality (Gao et al. 2023; Peng et al. 2023; Yamada 2023). For the field of literary translation, a number of studies also reported better results for GPT models compared to NMT, again highlighting the relevance of task and domain information in prompts. Karpinska and Iyyer (2023), for instance, found that ChatGPT-3.5 outperformed Google Translate when asked to translate paragraphs from a literary work rather than sentence-by-sentence (the study included eight source languages and three target languages, i.e., Polish, English, and Japanese). Wang et al. (2023) investigated the performance of LLMs and NMT for Chinese-English and English-German/ Russian for several text types, including web fiction, and also found LLMs to generate better results. In a study of translations of classical Chinese poetry into English by Gao et al. (2024), ChatGPT-3.5 outperformed DeepL and Google Translate in terms of fidelity, fluency, language style, and what the authors called the avoidance of “machine translation style”, especially when using prompts that contained task and domain information. Puppel and Borg (2025) also explored the impact of prompting (ChatGPT-3.5) on the translation of creative texts, in this case a travel blog from English into German. Interestingly, the best results were achieved with a simple zero-shot prompt (i.e., a prompt that did not include one or more translation examples serving as a reference), but all in all, the authors still found the text’s artistic style to be a great challenge, not least due to ChatGPT’s tendency to retain the source text’s syntactic structure. This is largely in line with findings in Du et al. (2025) that “less appears to be more when prompting ChatGPT to output a creative translation” (587), and ChatGPT-4o’s performance regarding creativity, based on different configurations in four languages, was also judged to be less successful than Human Translation (HT). Resende and Hadley (2024) compared HT with the performance of three LLMs (ChatGPT-3.5, ChatGPT-4, and Google Gemini) for poetry translation (Brazilian Portuguese/ Spanish-English) and also concluded that their output was not of a good enough quality (yet) to make sense as a draft for subsequent PE; given the opportunities offered by prompt engineering, however, they do recognize the systems’ potential of serving as a source of inspiration for translators, especially with a view to rhyme schemes. Another strand of research addresses the challenges of evaluating literary AT The automatic translation of literature and literary post-editing 227 <?page no="228"?> 3 https: / / deepl.com; https: / / openai.com. In July 2024, shortly after completion of the study, DeepL announced the introduction of an LLM for its translation service: https: / / www.deepl.com/ en/ blog/ next-gen-language-model#what_you_need_to_know_abou t_our_latest_language_model. output, with two recent studies looking into the advantages of LLM-based question-answer frameworks (Shafayat et al. 2025; Zhang et al. 2025), designed to outperform automatic evaluation metrics, such as BLEU. Research on literary PE has also gained momentum over the past few years, with much of it revolving around the related issues of creativity, translatorial voice, and priming. Guerberof-Arenas and Toral (2022), for example, found higher creativity scores in human translations than in NMT and PE versions for the language pairs English-Catalan/ Dutch. Winters and Kenny (2023) traced ways in which a German translator introduced his own translatorial voice into his post-edited version of a DeepL draft, while in an earlier study, they compared the same translator’s PE work on another English source text with a translation from scratch of that text that he had made years earlier, finding that his voice was less manifest in the PE version (Kenny and Winters 2020). Such constraints on translatorial voice and creativity in a literary PE task have been linked to priming effects induced by the AT draft. In a comparison of linguistic features of PE versions and human translations (English-Brazilian Portuguese), Castilho and Resende (2022), for example, see priming as a possible factor behind post-editese, the relative similarity between an MT draft and subsequent PE versions (see also Kolb 2023b and Kolb 2025 on NMT priming). While all of the above studies relied on NMT drafts, a more recent study by Castaldo et al. (2025) used three LLMs (GPT-4, GPT-3.5, and a literary-adapted Mistral-7B model); comparing the LLM-based post-edited versions with human translations of excerpts from a novel, the authors found that differences in creativity were minimal, and given significant reductions in task time, they see LLMs as a potentially more promising tool for literary PE than NMT. 3 Design of the Chopin study The main objectives of this study were 1) to compare the performance of an NMT system (DeepL Pro) with that of an LLM (ChatGPT-4 Turbo) 3 for an English-to-German literary translation task, and 2) to investigate how MA-level translation students post-edited the AT drafts. The Source Texts (ST) chosen for this study were two short stories by the 19 th -century American writer Kate Chopin (1850-1904). Chopin is best known for her short stories, which appeared in prestigious magazines and depicted scenes from the lives of women 228 Waltraud Kolb <?page no="229"?> 4 https: / / www.freescreenrecording.com/ . I thank the participants for taking on this extra task (for which they did not receive credits or grades) on top of their regular workload. in Louisiana at the end of the 19 th century. In “The Kiss”, first published in Vogue magazine in 1894, she portrays a woman who is intent on marrying a rich suitor even though she is secretly in love with another man; “The Recovery”, published two years later in the same magazine, tells us how a woman, after miraculously regaining her sight after fifteen years, is disillusioned with her changed physical appearance and life. As “The Kiss” and “The Recovery” only comprise 1,028 and 1,298 words, resp., the complete stories, rather than excerpts, could be used for the study. For reasons of space constraints, this contribution will limit itself to an analysis of the German target versions of “The Kiss”. The participants of the study were five female students and one male (all aged between 25 and 30) in the final stage of their two-year master’s program at the Center for Translation Studies of the University of Vienna. All of them were enrolled in a track of the program that focuses on the translation of literature, audiovisual texts, and the arts; all six had German as their first language and, with one exception, English as their main foreign language (the exception was participant P6 with French, who self-assessed her English skills at B2/ C1). The participants worked on the PE tasks at home, using the free screen-recording software Free Cam 4 to capture their processes, and filled out a questionnaire after completion of the task (a discussion of these processes, including the results of the questionnaires, will be the subject of a separate publication). Each participant post-edited a DeepL draft of one story and a ChatGPT draft of the other story so that for each story there are six PE versions available for analysis (three for each AT draft), and for each AT system there are also six PE versions available (three for each story). The participants did not generate the AT drafts themselves but were provided with the versions that I had generated in advance (April 2024; see also footnote 3). The prompt I used for ChatGPT was a zero-shot prompt that included the following domain and task information: “You are a literary translator. Please translate the following English text into German, using a temperature between 0.5 and 0.7 as I am interested in a balanced mix of creativity and coherence. Here is the English text to be translated into German: [source text]”. The setting of the temperature was inspired by previous research and prompting advice provided by ChatGPT (see, e.g., Jiao et al. 2023; Peng et al. 2023). The automatic translation of literature and literary post-editing 229 <?page no="230"?> 5 My thanks go to the eight literary translators for dedicating their time. 4 German versions of “The Kiss” In the following discussion of AT drafts and final PE versions of “The Kiss”, the focus will be on the product level, as indicated above. Participants of the study (hereinafter designated as P1 through P6) were asked to edit the AT drafts up until the point where they felt their target texts were of a publishable quality, good enough to be sent to a large German publisher of fiction for a new collection of US short stories from different time periods. The participants’ assessment of what constitutes publishable quality greatly varied - for instance, P1 edited approx. 30 % of the DeepL draft of “The Kiss”, while P2’s interventions amounted to just below 10 % of the same draft (see Table 1). High intersubject variation does not come as a surprise, though, and has also been found in studies involving professional translators (e.g., Kolb 2023b); in the case of translation students, differences can also be expected to arise from varying levels of linguistic and literary skills. Inspired by previous research (de Almeida and O’Brien 2010; Nitzke and Gros 2020), I will differentiate between essential edits and preferential edits (see also Kolb 2023a). In the context of this study, essential edits are edits of “AT errors”, i.e., AT suggestions which need to be rectified in order to achieve a target text that is correct both on the level of core meanings and the level of grammar and orthography. Preferential edits, on the other hand, go beyond the correction of such errors and are changes made by post-editors to further improve the target text, often reflecting personal preferences. Needless to say, determining the (in)correctness of a word or phrase (i.e., whether an AT suggestion constitutes an error or is just not a very good solution) is not as clear-cut and unequivocal in all cases as it might seem at first glance. To mitigate rater bias, a group of nine experienced literary translators, including myself, all members of the Austrian professional association of literary translators, put together a list of AT errors in the course of an in-person discussion 5 ; essential edits made by participants were then identified based on this list (see Table 2; for a particular item to count as an AT error, the majority of the group had to agree). Before addressing AT errors and the post-editors’ essential and preferential edits, the scope of edits made by the participants will be discussed; as indicated above, the perceived need of intervention greatly varied between post-editors. 230 Waltraud Kolb <?page no="231"?> 4.1 Edit distances Edit distance measures the (dis)similarity between an AT draft and a post-edited version of that draft and can therefore serve as a rough indicator of AT quality as it is perceived by a post-editor and the post-editor’s subsequent PE effort. For this study, CharCut was used to determine edit distances between the two AT drafts and the final PE versions; it is a character-based metric developed by Lardilleux and Lepage (2017) that calculates the number of operations (insertions, deletions, and shifts) required to transform one text into the other. For the purpose of this contribution, intermediate versions and “edits of edits” made during successive editing rounds are disregarded. In Table 1, higher CharCut scores mean higher numbers of edits; a score of, for instance, 0.2963, means that approx. 30 % of the AT draft were changed and approx. 70 % remained unchanged. For a better interpretation of results, Table 1 also includes the scores for the second story (“The Recovery”). PE version DeepL Kiss ChatGPT Kiss DeepL Recovery ChatGPT Recovery P1 0.2963 - - 0.2338 P2 0.0937 - - 0.0463 P3 0.1418 - - 0.1429 P4 - 0.1418 0.1339 - P5 - 0.1177 0.0458 - P6 - 0.0578 0.0307 - Average AT version 0.1773 0.1058 0.0701 0.1410 Average Kiss 0.1415 - Average Recovery - 0.1056 Average DeepL 0.1237 Average ChatGPT 0.1234 Tab. 1: Edit distances between DeepL/ ChatGPT drafts and final PE versions as well as averages for AT versions, stories and AT systems As shown in Table 1, edit distances are not consistently different between the two AT drafts. While in the case of “The Kiss”, the average edit distance is higher for DeepL than for ChatGPT (0.1773 vs. 0.1058), it is the other way around The automatic translation of literature and literary post-editing 231 <?page no="232"?> 6 https: / / www.maxqda.com/ de/ for “The Recovery” (0.0701 vs. 0.1410). With the exception of participant P3 (who made approx. the same number of edits in both stories), all participants made more edits in “The Kiss” (resulting in an average distance score of 0.1415) than in “The Recovery” (with an average score of 0.1056) - a possible explanation was put forth by P1 in her questionnaire, saying that she felt “The Kiss” took more effort to post-edit as it is more deeply rooted in socio-cultural context. Average scores across both stories for the two AT systems are almost identical: 0.1237 (DeepL) vs. 0.1234 (ChatGPT). However, edit distances greatly varied between participants, with P1 and P6 at the two opposite poles (0.2963 vs. 0.0578 for “The Kiss”; 0.2338 vs. 0.0307 for “The Recovery”). P1 clearly stands out as the most active post-editor (with scores of 0.2963/ 0.2338), with the other five participants ranging between 0.1418/ 0.1429 (P3) and 0.0578/ 0.0307 (P6). While P1’s level of intervention is roughly in line with findings from two of my studies involving professional literary translators (see Kolb 2023a, 2023b), the scores of the other five participants are considerably lower. In this study, the extent to which the post-editors found it necessary to intervene and edit the AT drafts was thus not so much determined by the AT systems used but rather by the different challenges posed by the two stories as well as different personal styles of the post-editors and especially, given the fact that participants were translation students rather than professionals, different skill levels. 4.2 AT errors and essential edits As described above, errors in the two AT drafts of “The Kiss” were determined by a group of nine experienced literary translators. Edits of these errors were then manually annotated by me using MAXQDA 6 . The assignment of error and edit categories was inspired by previous studies (Tezcan et al. 2019; Macken et al. 2022), with one significant difference: style is not treated here as a separate category; instead, I proceed from a notion of style as the overall linguistic and discoursal patterning of a text as the result of linguistic choices made by authors or, in our case, post-editors on all levels, be it the level of lexico-semantics or syntax or, let’s say, morphology. These choices are seen as “significant rather than merely ornamental”, i.e., content and form are seen here as “indivisible” (Mullany and Stockwell 2015, 45; see also Kolb 2023a). Table 2 provides an overview of AT errors and essential edits made by the post-editors. Given that the participants were translation students, it was also deemed of interest to determine whether they failed to spot some AT errors 232 Waltraud Kolb <?page no="233"?> and/ or unwittingly introduced new errors through their edits; both scores are therefore also included in Table 2. Errors of orthography in “The Kiss” exclusively relate to wrongly positioned opening quotation marks (which in German - unlike in English - should be placed at the bottom). DeepL failed to follow the German convention in all 19 cases, while ChatGPT made this mistake only twice; in order to not unduly distort results, the number of total errors in Table 2 disregards orthography/ quotation marks. AT error/ essential edit DeepL P1 P2 P3 Chat GPT P4 P5 P6 Semantics/ literalism 17 17 14 14 17 9 15 4 Semantics/ reference 13 13 13 13 10 10 9 5 Semantics/ other 12 11 3 5 7 5 4 3 Syntax 4 4 3 4 1 1 1 0 Morphology 4 4 1 3 3 2 3 1 [Orthogr.] [19] [18] [19] [19] [2] [0] [0] [0] Omission 1 1 0 1 2 1 2 1 Addition 0 0 0 0 0 0 0 0 Total (%) 51 50 (98) 34 (67) 40 (78) 40 28 (70) 34 (85) 14 (35) Missed errors incl. orthogr. - 2 17 11 - 14 8 28 New errors - 0 4 1 - 2 3 Tab. 2: AT errors, essential edits, missed errors, and newly introduced errors in “The Kiss” The vast majority of errors in both AT drafts relate to semantics, with two subtypes standing out more or less equally in DeepL and ChatGPT: errors due to literal translations and errors of reference. Unlike other studies (e.g., Raunak et al. 2023), this study of “The Kiss” - limited as it is - did not find ChatGPT to outperform NMT in terms of literalness (see below). Regarding other semantic errors, however, ChatGPT achieved better results than DeepL, as it did on the level of syntax. The generally low numbers of syntax and morphology errors The automatic translation of literature and literary post-editing 233 <?page no="234"?> attest to the progress AT systems have made over the past few years, as does the fact that none of the AT systems wrongly added any textual material that is not in the ST. In both AT drafts, however, we find several instances of omissions; proceeding from a functional perspective on translation (Nord 2012), omissions were only counted as errors here when they led to an incorrect target text, which occurred once in the DeepL version and twice in the ChatGPT draft: for example, translating the sentence “Brantain slowly arose; so did the girl arise, but quickly”, DeepL omitted the second mention of the verb arise, thereby generating a contradiction in terms. A slight tendency of ChatGPT to omit more ST material than DeepL occurred also with omissions that were not considered errors (five vs. three instances). As pointed out above, one of the main sources of semantic errors were literal translations (see also, e.g., Corpas Pastor and Noriega-Santiáñez 2024; Puppel and Borg 2025). In some cases, both systems failed at the same idioms: for instance, to retrace their steps was translated literally and inappropriately by both as ihre Schritte zurückverfolgen. Also, both systems sometimes used the next best or direct equivalent of a word, disregarding context and connotations. For example, in a scene in which the female protagonist Nathalie and Brantain, her visitor, are sitting in her drawing room, Chopin describes Brantain as Nathalie’s companion; both systems opted for the German term Begleiter, which is used for someone accompanying another person on a walk, a trip, or some other activity but not for someone coming to another person’s home for a visit (see, e.g., Iyer et al. 2023 on ambiguity and polysemy as a challenge for AT systems). One well-known problem when translating dialogues from English into German is the translation of the pronoun you. In German, the choice between the formal Sie and the informal du carries social and emotional meaning and depends on the setting of a story, the relationship between characters, social norms etc. In “The Kiss”, Nathalie and her suitor, Brantain, would be expected to address each other by Sie, given social conventions at the time the story unfolds, whereas Nathalie and her childhood friend Harvy, who is almost part of the family, would use the informal du. The two AT systems got approx. two thirds of the cases right (DeepL: 18, ChatGPT: 21, out of 31), the slightly better result for ChatGPT when it comes to reference being in line with other studies (e.g., Castilho et al. 2023). The choices of pronouns seem random, however, and both systems occasionally even used a mix of both pronouns within the same dialogue, which shows that cultural knowledge and social conventions are still very much the domain of human translators. When it comes to missed and newly introduced errors, participant P1 stands out with the best results (she spotted all AT errors with the exception of an 234 Waltraud Kolb <?page no="235"?> incorrect pronoun and a wrong quotation mark and did not introduce any new errors). Higher scores for the other participants indicate that, as students, they may still be insecure regarding some linguistic choices and more susceptible to AT priming effects (see Kolb 2023b for NMT priming). 4.3 Preferential edits Table 3 shows the preferential edits made by the six participants. The total number of preferential edits is approx. 37 % lower in the ChatGPT draft than in the DeepL version. However, if we take a closer look at the results we see that this difference is mainly due to participant P1 who was by far the most active post-editor, making a total of 202 edits in the DeepL draft, more than twice the number of the participant who comes next in terms of edit numbers (P4, 95 edits in the ChatGPT draft). If we disregard P1, numbers of preferential edits for the other five participants range from 38 to 95 and are distributed more or less equally across the two AT drafts. The vast majority of edits made by each participant were of a preferential nature - in P1’s case 80 % of all edits were preferential, and even with the two least active post-editors (P2 and P6), preferential edits amounted to 59 % and 73 %, resp. The average of 73 % is the same for DeepL and ChatGPT. As in the case of essential edits, the participants introduced some new errors through their edits. Not surprisingly, some of these new errors are also linked to the fact that the ST is over a hundred years old: for instance, all three errors introduced by P1 relate to register, such as the too modern and colloquial expression Klatsch und Tratsch (idle gossip) for tea. The automatic translation of literature and literary post-editing 235 <?page no="236"?> DeepL ChatGPT Preferential edit P1 P2 P3 Total P4 P5 P6 Total Semantics 103 30 32 165 45 34 21 100 Syntax 42 14 27 83 27 20 6 53 Morphology 9 0 3 12 4 7 2 13 Orthogr. 0 0 0 0 0 0 0 0 Deletion 16 1 8 25 2 8 5 15 Addition 32 3 14 49 17 8 4 29 Total (%) 202 (80) 48 (59) 84 (68) 334 (73) 95 (77) 77 (69) 38 (73) 210 (73) New errors 3 4 0 7 8 4 1 13 Tab. 3: Preferential edits, percentage of total edits, and newly introduced errors in “The Kiss” As was the case with essential edits, the largest group is again semantics, across both AT drafts and all participants. The number of syntactical edits is relatively high, even though both systems hardly made any syntax errors (see also Kolb 2023a; Puppel and Borg 2025). We can take this to indicate that even if syntactical structures in AT drafts are not incorrect, they are often perceived as not quite natural and in need of improvement. What Gao et al. (2024) have described as “machine translation style” can be mitigated by post-editors by restructuring sentences, but also, for example, by adding modal particles in a German target text. Indeed, many of the additions made by participants (P1 made a total of 31 additions, the other participants between three and 17) are modal particles, which are much more frequently used in German than in English. On the level of semantics, they add nuance, emotion, or emphasis or subtly convey attitudes or assumptions without changing the core meaning of a sentence; on the syntactical level, their almost complete absence in the AT outputs arguably has an effect of abruptness or harshness on readers. In the following example from P1’s final PE version, she added three modal particles (wohl/ surely; schon/ already; ja/ obviously, after all) and an adverb (jetzt/ now) to the DeepL draft (all four additions in italics below). She also made some other edits, which, however, have no bearing on the use of modal particles here and are not our concern at this point: 236 Waltraud Kolb <?page no="237"?> “I believe,” stammered Brantain, “I see that I have stayed too long. I--I had no idea--that is, I must wish you good-by.” (ST) “Ich glaube”, stammelte Brantain, “ich sehe, dass ich zu lange geblieben bin. Ich ich hatte keine Ahnung das heißt, ich muss Ihnen Lebewohl sagen.“ (DeepL) „Ich glaube“, stammelte Brantain, „ich merke, dass ich wohl schon zu lange hier gewesen bin. Ich ich wusste ja nicht - ich meine, ich muss mich jetzt verabschieden." (P1) Both DeepL and ChatGPT sometimes used normalizing language, translating ST words that are foregrounded and carry specific meanings and connotations with more general and high-frequency target terms (Vanmassenhove et al. 2019); such cases were not considered errors, and any edits were thus counted as preferential. An example is ardent(ly), which occurs twice in the ST, each time in the context of Nathalie’s relationship with the two male protagonists: we learn how Brantain, the suitor who Nathalie intends to marry for his money but does not find attractive, keeps his eyes fastened ardently upon her in the drawing room scene, and Harvy, Nathalie’s childhood friend, whom she is in love with, pressed an ardent, lingering kiss upon her lips. Other words from the same semantic field used to describe the scene, such as smouldering fire and firelight, further contribute to this effect of foregrounding (see also Tearle 2023). For both instances of ardent, the two AT systems used identical target words: intensiv (intense) in the first case and leidenschaftlich (passionate) in the second (approx. 78 % of all adjectives were identical in both AT drafts). Both terms are high-frequency everyday German words that are much paler compared to ardent(ly). The post-editors failed to recognize that Chopin here made use of the stylistic devices of foregrounding and repetition to subtly convey the cognitive and emotional state of the protagonists: all participants retained leidenschaftlich, and only three of them replaced intensiv with another adverb, though in no case with one coming close to achieving a comparable effect to the ST term. 5 Conclusion This contribution discussed some findings from a study that explored how two AT systems, DeepL and ChatGPT, translated two short stories written by US author Kate Chopin and how the AT drafts were post-edited by translation students. A number of previous studies have found LLMs to outperform NMT also for literary translation; this could not be fully confirmed here. If we look at edit distances as an indicator of perceived AT quality, we find that the average scores across both stories are almost identical for both AT systems. For “The Kiss”, the story which was the focus of this contribution, ChatGPT The automatic translation of literature and literary post-editing 237 <?page no="238"?> did generate fewer errors on the level of syntax and regarding some aspects of semantics, but not, for instance, when it comes to literalism, one of the great challenges for any AT system. Regarding preferential edits, which went beyond the amendment of AT errors and were made to further improve the target text, results are comparable for the two systems, if we disregard participant P1, who was an outlier as the by far most active post-editor. The great majority of preferential edits in both AT drafts concerned semantics and syntax, relating again, in particular, to literal AT suggestions but also, for instance, to register and language use from more than a century ago (see also Cespedosa Vázquez and Mitkov 2023). The participants of this study were six translation students in the last phase of their MA studies. Their work shows a high degree of intersubject variation, which can be attributed to different personal preferences but also to different levels of linguistic skills and confidence, with the linguistic sensibility of one participant (P1) being clearly more advanced than that of her colleagues. Need‐ less to say, results would certainly have turned out somewhat different had the participants been experienced literary translators - in which case higher num‐ bers of preferential edits could have been expected. Further studies are needed that involve larger groups of participants, additional AT systems, language pairs, and types of literature. The presented findings therefore should be interpreted with caution; what they certainly show, however, is that any “meaningful use of machine translation” (Budin 2019: 54) for literary texts, regardless of the AT system used, inevitably involves careful PE and that literary PE - just like translation from scratch - is a highly demanding task that requires very advanced linguistic and literary skills. Bibliography Budin, Gerhard (2019). Die Maschinelle Übersetzung - 70 Jahre jung! In: Günther, Johann (Ed.). Veränderungen in einer Generation: Beiträge aus Wissenschaft, Kunst, Philosophie, Literatur und Management. Wien: Freya, 49-55. Castaldo, Antonio/ Castilho, Sheila/ Moorkens, Joss/ Monti, Johanna (2025). Extending CREAMT: Leveraging Large Language Models for Literary Translation Post-Edi‐ ting. In: Bouillon, Pierrette/ Gerlech, Johanna/ Girletti, Sabrina/ Volkart, Lise/ Rubino, Raphael/ Sennrich, Rico/ Farinha, Ana C./ Gaido, Marco/ Daems, Joke/ Kenny, Doro‐ thy/ Moniz, Helena/ Szoc, Sara (Eds.). Proceedings of Machine Translation Summit XX, Volume 1. Geneva: European Association for Machine Translation, 506-515. Castilho, Sheila/ Resende, Nátalia (2022). Post-editese in Literary Translations. Informa‐ tion 13 (2), 66. 238 Waltraud Kolb <?page no="239"?> Castilho, Sheila/ Quinn Mallon, Clodagh/ Meister, Rahel/ Yue, Shengya (2023). Do Online Machine Translation Systems Care for Context? What About a GPT Model? In: Nurminen, Mary/ Brenner, Judith/ Koponen, Maarit/ Latomaa, Sirkku/ Mikhailov, Mik‐ hail/ Schierl, Frederike/ Ranasinghe, Tharindu/ Vanmassenhove, Eva/ Alvarez Vidal, Sergi/ Aranberri, Nora/ Nunziatini, Mara/ Parra Escartín, Carla/ Forcada, Mikel/ Popo‐ vic, Maja/ Scarton, Carolina/ Moniz, Helena (Eds.). Proceedings of the 24th Annual Conference of the European Association for Machine Translation. Tampere: European Association for Machine Translation, 393-417. Cespedosa Vázquez, Ana Isabel/ Mitkov, Ruslan (2023). Machine Translation of Literary Texts: Genres, Times and Systems. In: Lázaro Gutiérrez, Raquel/ Escriche, Marie/ Mi‐ hajlov, Teodora/ Kunilovskaya, Maria/ Mitkov, Ruslan (Eds.). Proceedings of the First Workshop on NLP Tools and Resources for Translation and Interpreting Applications. Varna: INCOMA Ltd., 48-53. Chopin, Kate (1896). The Recovery. Available at: https: / / americanliterature.com/ author/ kate-chopin/ short-story/ the-recovery/ (Accessed: 01.04.2026). Chopin, Kate (1894). The Kiss. Available at: https: / / americanliterature.com/ author/ kate -chopin/ short-story/ the-kiss/ (Accessed: 01.04.2026). Corpas Pastor, Gloria/ Noriega-Santiáñez, Laura (2024). Human versus Neural Machine Translation Creativity: A Study on Manipulated MWEs in Literature. Information 15, 530. De Almeida, Giselle/ O’Brien, Sharon (2010). Analysing post-editing performance: Corre‐ lations with years of translation experience. In: Toral, Antonio/ Castilho, Sheila (Eds.). Proceedings of the 14th Annual Conference of the European Association for Machine Translation. Saint Raphaël, France: European Association for Machine Translation 27-28. Du, Shuxiang/ Guerberof-Arenas, Ana/ Toral, Antonio/ Gerrits, Kyo/ Marco Borillo, Josep (2025). Optimising ChatGPT for creativity in literary translation: A case study from English into Dutch, Chinese, Catalan and Spanish. In: Way, Andy/ O’Brien, Sharon/ Moorkens, Joss (Eds.). Proceedings of Machine Translation Summit XX, Volume 1. Geneva: European Association for Machine Translation, 578-591. Gao, Ruiyao/ Lin, Yumeng/ Zhao, Nan/ Cai, Zhenguang G. (2024). Machine translation of Chinese classical poetry: a comparison among ChatGPT, Google Translate, and DeepL Translator. Humanities and Social Sciences Communications 11, 835. Gao, Yuan/ Wang, Ruili/ Hou, Feng (2023). How to design translation prompts for ChatGPT: an empirical study. CoRR abs/ 2304.02182. Genzel, Dmitriy/ Uszkoreit, Jakob/ Och, Franz (2019). „Poetic“ Statistical Machine Trans‐ lation: Rhyme and Meter. In: Yamada, Masaru/ Korhonen, Anna (Eds.). Proceedings of the 2010 Conference on Empirical Methods in Natural Language Processing. MIT, Massachusetts: Association for Computational Linguistics, 158-166. The automatic translation of literature and literary post-editing 239 <?page no="240"?> Guerberof-Arenas, Ana/ Toral, Antonio (2022). Creativity in translation: Machine trans‐ lation as a constraint for literary texts. Translation Spaces 11 (2), 184-212. Hansen, Damien/ Esperança-Rodier, Emmanuelle/ Blanchon, Hervé/ Bada, Valérie (2022). La traduction littéraire automatique: Adapter la machine à la traduction humaine individualisée. Journal of Data Mining and Digital Humanities: Vers une robotique du traduire, Towards robotic translation? https: / / hal.science/ hal-03583562 (Accessed: 01 .04.2026). Hendy, Amr/ Abdelrehim, Mohamed/ Sharaf, Amr/ Raunak, Vikas/ Gabr, Mohamed/ Mat‐ sushita, Hitokazu et al. (2023). How good are gpt models at machine translation? a comprehensive evaluation. CoRR abs/ 2302.09210. Iyer, Vivek/ Chen, Pinzhen/ Birch, Alexandra (2023). Towards effective disambiguation for machine translation with large language models. In: Philipp Koehn/ Barry Had‐ dow/ Tom Kocmi/ Christof Monz (Eds.). Proceedings of the Eighth Conference on Machine Translation. Singapore: Association for Computational Linguistics, 482-495. Jiao, Wenxiang/ Wang, Wenxuan/ Huang, Jen-tse/ Wang, Xing/ Shi, Shuming/ Tu, Zhao‐ peng (2023). Is ChatGPT a good translator? yes with GPT-4 as the engine. CoRR abs/ 2301.08745 Karpinska, Marzena/ Iyyer, Mohit (2023). Large language models effectively leverage document-level context for literary translation, but critical errors persist. In: Philipp Koehn/ Barry Haddow/ Tom Kocmi/ Christof Monz (Eds.). Proceedings of the Eighth Conference on Machine Translation. Singapore: Association for Computational Lin‐ guistics, 419-451. Kenny, Dorothy/ Winters, Marion (2020). Machine translation: ethics and the literary translator’s voice. Translation Spaces 9 (1), 123-149. Kolb, Waltraud (2023a). Brazilian short prose in German: a study of literary post-editing. Revista tradumàtica 21, 63-86. Kolb, Waltraud (2023b). I am a bit surprised: literary translation and post-editing processes compared. In: Andrew Rothwell/ Andy Way/ Roy Youdale (Eds.). Compu‐ ter-assisted literary translation: the state of the art. London: Routledge, 45-68. Kolb, Waltraud (2025). Creativity in literary post-editing: reworking a German DeepL translation of a text by Lima Barreto. In: Katharina Walter/ Macro Agnetto (Eds.). Applying artificial intelligence in translation and interpreting: possibilities, processes and phenomena. London: Routledge, 99-118. Kuzman, Taja/ Vintar, Špela/ Arčan, Mihael (2019). Neural machine translation of literary texts from English to Slovene. In: Sheila Castilho/ Joss Moorkens (Eds.). Proceedings of the Qualities of Literary Machine Translation. Dublin: European Association for Machine Translation, 1-9. Lardilleux, Adrien/ Lepage, Yves (2017). CHARCUT: human-targeted character-based MT evaluation with loose differences. In: Kevin Knight/ Marcello Federico (Eds.). 240 Waltraud Kolb <?page no="241"?> Proceedings of the 14th International Workshop on Spoken Language Translation. Tokyo: Association for Computational Linguistics, 146-153. MacDonald, Keza (2024). It gets more and more confused: can I replace translators? The Guardian, 12. November. https: / / www.pressreader.com/ australia/ the-guardian-austra lia/ 20241112/ 281947433374511 (Accessed: 01.04.2026). Macken, Lieve/ Vanroy, Bram/ Desmet, Luca/ Tezcan, Arda (2022). Literary translation as a three-stage process: machine translation, post-editing and revision. In: Sheila Castilho/ Joss Moorkens (Eds.). Proceedings of the 23rd Annual Conference of the European Association for Machine Translation. Ghent: European Association for Machine Translation, 101-110. Manakhimova, Shushen/ Avramidis, Eleftherios/ Macketanz, Vivien/ Lapshinova-Koltun‐ ski, Ekaterina/ Bagdasarov, Sergei/ Möller, Sebastian (2023). Linguistically motivated evaluation of the 2023 state-of-the-art machine translation: can GPT-4 outperform NMT? In: Philipp Koehn/ Barry Haddow/ Tom Kocmi/ Christof Monz (Eds.). Procee‐ dings of the Eighth Conference on Machine Translation. Singapore: Association for Computational Linguistics, 224-245. Mullany, Louise/ Stockwell, Peter (2015). Introducing English language: a resource book for students. London: Routledge. Nitzke, Jean/ Gros, Anne-Kathrin (2020). Preferential changes in revision and post-edi‐ ting. In: Maarit Koponen/ Brian Mossop/ Isabelle S. Robert/ Giovanna Scocchera (Eds.). Translation revision and post-editing. Industry practices and cognitive processes. London: Routledge, 87-102. Nord, Christiane (2012). Quo vadis, functional translatology? Target 24 (1), 25-41. OMC (2022). Translators on the cover. Multilingualism & translation. Report of the Open Method of Coordination (OMC) Group of EU Member States’ Experts. Luxembourg: Publications Office of the European Union. Peng, Keqin/ Ding, Liang/ Zhong, Qihuang/ Shen, Li/ Liu, Xuebo/ Zhang, Min et al. (2023). Towards making the most of ChatGPT for machine translation. CoRR abs/ 2303.13780. Puppel, Melissa/ Borg, Claudine (2025). Evaluating ChatGPT’s performance in creative text translation for communication: a case study from English into German. Media and Intercultural Communication: A Multidisciplinary Journal 3 (1), 1-27. Raunak, Vikas/ Menezes, Arul/ Post, Matt/ Awadalla, Hany Hassan (2023). Do GPTs pro‐ duce less literal translations? CoRR abs/ 2305.16806. Resende, Natália/ Hadley, James (2024). The Translator’s Canvas: Using LLMs to Enh‐ ance Poetry Translation. In: Knowles, Rebecca/ Eriguchi, Akiko/ Goel, Shivali (Eds.). Proceedings of the 16th Conference of the Association for Machine Translation in the Americas, Volume 1: Research Track. Chicago, USA: Association for Machine Translation in the Americas, 178-189. The automatic translation of literature and literary post-editing 241 <?page no="242"?> Shafayat, Sheikh/ Yoon, Dongkeun/ Jang, Woori/ Choi, Jiwoo/ Oh, Alice/ Jung, Seohyon (2025). A 2-step framework for automated literary translation evaluation: its promises and pitfalls. CoRR abs/ 2412.01340. Tearle, Oliver (2023). Summary and analysis of Kate Chopin’s The Kiss. https: / / interest ingliterature.com/ 2023/ 08/ kate-chopin-the-kiss-summary-analysis/ (Accessed: 01.04. 2026). Tezcan, Arda/ Daems, Joke/ Macken, Lieve (2019). When a sport is a person and other issues for NMT of novels. In: Sheila Castilho/ Joss Moorkens (Eds.). Proceedings of the Qualities of Literary Machine Translation. Dublin: European Association for Machine Translation, 40-49. Toral, Antonio/ Way, Andy (2018). What level of quality can neural machine translation attain on literary text? In: Joss Moorkens/ Sheila Castilho/ Federico Gaspari/ Stephen Doherty (Eds.). Translation quality assessment. From principles to practice. Cham: Springer, 171-190. Vanmassenhove, Eva/ Shterionov, Dimitar/ Way, Andy (2019). Lost in translation: loss and decay of linguistic richness in machine translation. In: Nicoletta Calzolari/ Kate Walker/ Marcello Federico (Eds.). Proceedings of MT Summit XVII: Research Track. Nice: International Association for Machine Translation, 222-232. Wang, Longyue/ Lyu, Chenyang/ Ji, Tianbo/ Zhang, Zhirui/ Yu, Dian/ Shi, Shuming et al. (2023). Document-level machine translation with large language models. CoRR abs/ 2304.02210. Webster, Rebecca/ Fonteyne, Margot/ Tezcan, Arda/ Macken, Lieve/ Daems, Joke (2020). Gutenberg goes neural: comparing features of Dutch human translations with raw neural machine translation outputs in a corpus of English literary classics. Informatics 7 (3). Winters, Marion/ Kenny, Dorothy (2023). Mark my words: a translator-specific explora‐ tion of style in literary machine translation. In: Andrew Rothwell/ Andy Way/ Roy Youdale (Eds.). Computer-assisted literary translation: the state of the art. London: Routledge, 69-92. Yamada, Masaru (2023). Optimizing Machine Translation through Prompt Engineering: An Investigation into ChatGPT’s Customizability. In: Yamada, Masaru/ do Carmo, Félix (Eds.). Proceedings of Machine Translation Summit XIX, Volume 2: Users Track. Dublin: Association for Machine Translation in the Americas, 195-204. Yirmibeşoğlu, Zeynep/ Dursun, Olgun/ Dallı, Harun/ Şahin, Mehmet/ Hodzik, Ena/ Gürses, Sabri et al. (2023). Incorporating human translator style into English-Turkish literary machine translation. In: Sheila Castilho/ Joss Moorkens (Eds.). Proceedings of the 24th Annual Conference of the European Association for Machine Translation. Tampere: European Association for Machine Translation, 419-428. 242 Waltraud Kolb <?page no="243"?> Zhang, Ran/ Zhao, Wei/ Eger, Steffen (2025). How Good Are LLMs for Literary Translation, Really? Literary Translation Evaluation with Humans and LLMs. In: Chiruzzo, Luis/ Ritter, Alan/ Wang, Lu (Eds.). Proceedings of the 2025 Conference of the Nations of the Americas Chapter of the Association for Computational Linguistics: Human Language Technologies, Volume 1: Long Papers. Albuquerque, New Mexico: Association for Computational Linguistics, 10961-10988. The automatic translation of literature and literary post-editing 243 <?page no="245"?> Multilingualism and Video Game Localisation Xiaochun Zhang Abstract: Video games have become a global cultural force, shaping enter‐ tainment, learning, and social interaction through immersive storytelling and player agency. Their creative and cognitive benefits, combined with their economic impact, position gaming as a defining industry of the digital age. At the same time, multilingualism, both individual and societal, plays a crucial role in shaping communication and identity in an increasingly interconnected world. Translation sits at the centre of this relationship, mediating linguistic diversity across media. While multilingualism has been widely studied, its presence in video games and its implications for localisation remain underexplored. This chapter examines how multilin‐ gualism enriches narrative depth, characterisation, and worldbuilding in games, and how localisation practices adapt these multilingual elements for diverse linguistic and cultural audiences, enhancing accessibility and player engagement. 1 Introduction Video games have become a worldwide influence, transforming entertainment, learning, and social connections. They offer immersive storytelling, allowing players to engage directly with narratives in ways traditional media cannot. Video gaming fosters creativity, problemsolving, and strategic thinking, making them a valuable tool for learning (Newman 2004, 2013). Games also connect people across cultures and languages, building communities where players collaborate and share experiences. Economically, gaming generates billions an‐ nually and continues to expand as a dominant force in media and technology. As gaming evolves, its influence on culture, communication, and global economies remains undeniable, solidifying its role as one of the most impactful industries of the digital age. <?page no="246"?> Multilingualism can be understood as either an individual’s characteristic or as a societal phenomenon. Individual multilingualism relates to “the perso‐ nal sphere and covers the acquisition and use of several languages by an individual” (Aronin 2019: 4). It pertains to a person’s capacity to acquire and effectively utilise multiple languages, encompassing both linguistic skills and the brain’s neurological processes. Societal multilingualism refers to “the contexts, circumstances, order, manner and routines of use of languages in different kinds of communities, organizations and groups” (ibid). The presence of societal multilingualism in a country or region does not imply that all its citizens are equally multilingual. For example, as explained by the Council of Europe (2007: 17), multilingualism is the presence of several languages in a given space, independently of those who use them: for example, the fact that two languages are present in the same geographical area does not indicate whether inhabitants know both languages, or only one. In many places, multiple languages coexist within communities and territories, yet individual language proficiency varies widely. It is both practical and logical to differentiate between individual and societal multilingualism. However, these domains are deeply interconnected rather than strictly separate (Aronin 2019: 4). Language, as Aronin rightly points out, is inherently a collective pheno‐ menon, making it impossible to examine individual multilingualism without acknowledging its broader societal influences. Likewise, understanding societal multilingualism requires an awareness of how multilingual experiences shape individuals (ibid.). Multilingualism is nowadays widespread across the globe, shaping commu‐ nication, culture, and education in diverse societies. The world is home to approximately 6,000 to 7,000 languages, spoken by a global population of 7 billion across 189 sovereign nations (Council of Europe 2015). There are approximately 3.3 billion bilingual individuals worldwide, representing 43 % of the global population (Gration 2025). Many multilingual individuals are immigrants who retain their first language(s) while integrating the languages of their host countries. Others acquire languages as they spread globally, often driven by economic and social opportunities (Cenoz 2013: 3). English, as the most widely spoken language, plays a significant role in this process, serving as a common subject in schools and a primary medium of instruction in educational institutions worldwide (e.g. Kirkpatrick 2011; Rao 2019). When multilingualism is understood as the coexistence of two or more languages within a text, individual, or society, and translation is conventionally defined as the replacement of one language with another, then translation and 246 Xiaochun Zhang <?page no="247"?> multilingualism are fundamentally interconnected, shaping and influencing each other in profound ways. As Meylaerts (2012: 227) aptly states, “at the heart of multilingualism, we find translation”. In Translation Studies, trans‐ lation has traditionally been understood as the process of transforming a monolingual source into a monolingual target product designed for a singlelan‐ guage audience. Since the late 1980s and early 1990s, descriptively oriented literary translation studies have recognised the necessity of shifting the focus toward the study of multilingualism and translation within broader social contexts, including media and other domains (Meylaerts and Serban 2014: 2). The intricate relationship between multilingualism and translation has only recently garnered significant attention in Translation Studies. Meylaerts (2006: 3) argues that Translation Studies is uniquely suited to exploring this field, as it provides a functional, descriptive perspective on issues of nationhood, sociopolitical dynamics, linguistic interactions, and cultural power struggles in multilingual societies. Research on this topic spans a wide range of methodolo‐ gies, disciplines, and areas of study. However, the intersection of multilingualism and video games, particularly its influence on game localisation, remains an underexplored area of study. This chapter explores the presence of multilingualism in video games and its influence on narrative depth, player interaction, and linguistic accessibility. Beyond this, the article delves into the intricate process of localisation, high‐ lighting how games are adapted for different linguistic and cultural audiences. It focuses on how onscreen multilingualism embraces linguistic diversity, enri‐ ching storytelling, character development, and worldbuilding. By incorporating multiple languages, games become more immersive and relatable, fostering deeper connections with players across different cultures. 2 Multilingualism and video games Video games have transcended mere entertainment, becoming a fundamental part of the digital landscape and captivating millions of players and enthusiasts worldwide. Video games are a distinct form of entertainment that sets itself apart from other media through interactivity. Unlike films, music, or television, which offer passive engagement, video games require active participation, immersing players in dynamic worlds where their actions generate immediate feedback and shape the unfolding experience in real time. Players assume the role of protagonists, directly influencing the narrative and even altering the course of a game’s storyline and ending, an aspect that enhances engagement and deepens immersion (Egenfeldt-Nielsen et al. 2019; Paterson et al. 2019). This Multilingualism and Video Game Localisation 247 <?page no="248"?> interactive nature allows players to connect with characters, explore different roles, and craft unique stories, particularly in role-playing games. This section examines how multilingualism intersects with video games, emphasising the role of multilingual elements in shaping immersive and interactive storytelling. 2.1 Multilingual people in the gaming industry Mass migration reshapes the linguistic landscape of societies worldwide. Ac‐ cording to the United Nations Department of Economic and Social Affairs (UNDESA), the global number of international migrants reached 304 million in 2024, accounting for 3.7% of the world’s population (United Nations 2025). This movement has contributed to the vast linguistic diversity of modern cities. London, for example, stands out as a prime example of multiculturalism, with over 300 languages spoken across the capital. Around 22 % of Londoners use a primary language other than English in their daily lives, and thousands of residents are multilingual, utilising multiple languages for social interactions and professional settings. Particularly notable is the younger generation, an estimated 41 % of state school children speak an additional language alongside English ( Jenkins 2022). In today’s interconnected world, individuals fluidly navigate multiple languages across various aspects of life, from home and work to social gatherings and online spaces. Modern nomadic citizens embody a constant state of (self)translation, seamlessly shifting between languages as they engage with diverse linguistic and cultural environments (Meylaerts 2013: 522). Multilingual societies provide an ideal environment for multilingual writers and selftranslators, fostering linguistic fluidity and creative expression across languages. For example, multilingualism and translation play a fundamental role in the collaborative filmmaking process, where directors, actors, and technical crews often come together from diverse linguistic and cultural backgrounds. These elements facilitate communication, creative expression, and seamless coordination across international teams, making them essential to the produc‐ tion of many films (Meylaerts and Serban 2014: 2). The gaming industry also exemplifies this dynamic, with its highly diverse and multicultural workforce. Video games are developed by teams with diverse cultural and linguistic backgrounds, contributing to the richness of their design and storytelling. According to the 2022 UK Games Industry Census, published by The Association for UK Interactive Entertainment (Ukie), nearly 30 % of professionals in the UK gaming industry hold a nationality other than British. This figure far exceeds the 11 % recorded for the broader UK workingage population, highlighting the industry’s global reach and inclusivity (Taylor 2022). Collaboration is the 248 Xiaochun Zhang <?page no="249"?> backbone of game development, with teams working across disciplines to bring concepts to life. Ruggill et al. (2016) identify six key areas: programming, art, design, producing, quality assurance, audio, and business, each playing a vital role in shaping a game’s final form. Major international gaming companies, such as Ubisoft, operate in 31 countries and employ teams representing 113 different nationalities. As stated on Ubisoft’s company website, this diversity enables them to better understand players and successfully adapt their games to a variety of cultural contexts (Ubisoft 2025). Their Assassin’s Creed franchise reflects this commitment to inclusivity, as the series has historically acknowledged its diverse development teams with an opening title card. For instance, Assassin’s Creed: Syndicate (Ubisoft 2015) includes the statement: Inspired by historical events and characters, this work of fiction was designed, deve‐ loped, and produced by a multicultural team of various beliefs, sexual orientations, and gender identities. Academic literature also suggests that multilingualism positively impacts work‐ places, as Houška (2025) demonstrates through an analysis of Czech video game studios, showing how diverse language strategies, beyond reliance on English, enhance inclusivity and collaboration in game development. 2.2 Multilingual texts in video games Multilingualism is deeply woven into various artistic forms, including literature, film, theatre, music, comics, and performance art, where language functions as both a narrative and aesthetic device. Writers have long used multilingualism to shape narratives, develop characters, and recreate historical contexts, reflecting the linguistic interactions of everyday life. By integrating multiple languages, literature captures the complexities of cultural and social dynamics, highlighting hierarchies and the fluidity of communication across linguistic landscapes (Monticelli et al. 2023: 7). This linguistic diversity appears in texts through single words, entire passages, dialects, sociolects, and foreign languages, serving multiple functions, enhancing plot development, shaping character dialogue, and reinforcing realism in storytelling (Meylaerts 2012: 227). Multilingualism manifests in game texts in various ways. Firstly, video games are typically developed by teams with diverse backgrounds, incorporating multiple text types. There are four primary categories of translatable assets: ingame text (e.g., menus and item descriptions), art assets (e.g., embedded text in graphics), audio and cinematic elements (e.g., cutscenes and trailers), and online materials (e.g., websites and social media promotions) (O’Hagan and Multilingualism and Video Game Localisation 249 <?page no="250"?> Mangiron 2013: 163-164). While game narratives are often crafted by designers or writers, technical texts such as menus and descriptions may be authored by programmers, and marketing content, including promotional materials and social media posts, is typically created by marketing teams. Additionally, many games are developed in a language that is not the native tongue of the creators. NonEnglishspeaking game studios often produce games in English, with nonfirst language Englishspeaking writers contributing to the scripts. During this process, translation, and sometimes selftranslation, often occurs in the story writing phase. Importantly, the cultural identity of a game’s source text does not always align with the locale of its development. As BernalMerino (2009: 243) highlights, a developer’s nationality should be inconsequential; what matters is that players feel the game was created with their cultural preferences in mind. Moreover, multilingual elements are a common feature in video games, with many titles seamlessly integrating multiple languages into their narratives to enhance authenticity and explore themes of cultural exchange and identity. A notable example is Assassin’s Creed, a historical actionadventure series that integrates languages such as Latin and Greek to reinforce the realism of its settings (Gainsford 2019). Ezio Auditore da Firenze, a fictional Italian master assassin, serves as the principal protagonist in the games set during the Italian Renaissance. His first language is Italian, but he is also proficient in French, has a strong grasp of Latin, and knows some Spanish. In Assassin’s Creed: Revelations (Ubisoft 2011), he even uses a few words in Greek and Turkish, further highlighting the game’s multilingual elements. Another example is Lara Croft in Tomb Raider (Embracer Group, 1996-2025), who showcases impressive multilingual abilities throughout the series. Her proficiency in over a dozen languages adds authenticity to her role as a globally knowledgeable adventurer, reinforcing the game’s themes of cultural exchange and identity. Whether translating ancient inscriptions on the fly or conversing fluently in Yucatec Mayan in Shadow of the Tomb Raider (EidosMontréal, Nixxes Software 2018), Lara’s linguistic skills play a crucial role in gameplay and storytelling. In Tomb Raider: Legend, she even surprises Russian soldiers by responding in their language, highlighting how multilingualism can shape character interactions in video games. There are several other games that feature multilingual characters, such as Tidus in Final Fantasy X (Square Enix 2001) who can become fluent in the fictional Al Bhed language through ingame learning; and Agent 47 in Hitman (IO Interactive 2021-present), a highly skilled assassin, who speaks multiple languages to blend into different environments, to name but a few 250 Xiaochun Zhang <?page no="251"?> (Mubeen 2025). These characters showcase how language diversity can enrich storytelling and worldbuilding. In addition, most games, particularly AAA titles are “born translated” (Zhang and Song 2023). Literary works that are “written for translation from the start” are described by Walkowitz (2015: 3) as “born translated”, meaning that translation is approached as an inherent medium rather than an afterthought. In this sense, most video games are developed with global distribution in mind, with localisation planned from the outset. Borntranslated works are written for translation, in the hope of being translated, but they are also often written as translations, pretending to take place in a language other than the one in which they have, in fact, been composed (Walkowitz 2015: 4). The Dark Souls series provides a pertinent example, as it is developed in Japan yet lacks overtly identifiable markers of Japanese cultural or aesthetic tradition. To both international and domestic players, the game may not immediately appear to be a Japanese video game. The main menu UI in the Japanese version is presented in English, while the medieval Europeaninspired setting is distinct from Japanese cultural aesthetics. The characters lack recognisably Japanese names, and most notably, there is no voiced dialogue in Japanese. In this way, the Japanese source text is deliberately crafted to resemble a translation of the English version (Zhang and Song 2023). Second, in borntranslated literature, translation functions “as a thematic, structural, conceptual, and sometimes even typographical device,” serving as “a spur to literary innovation” (Walkowitz 2015: 4). This is particularly relevant to the creative process of Hidetaka Miyazaki, the lead designer of Dark Souls. He drew inspiration from a variety of sources, with British fantasy literature, particularly the works of J.R.R. Tolkien, playing a crucial role. Some of the game’s creatures bear similarities to those found in The Lord of the Rings (Olivares 2021), and Miyazaki has described his approach as “cowriting the fiction” alongside the author (Parkin 2015). Translation is woven into the game’s aesthetic design, including location names such as Anor Londo ( アノール・ロン ド ) and Irithyll of the Boreal Valley ( 冷たい谷のイルシール ), which are derived from Tolkien’s Elvish. Additionally, many characters have Europeanorigin names, such as the highest deity, グウィン (Gūin), adapted from the Welsh word Gwyn, meaning white. To emphasise this reference, the game’s boss battle theme music was composed in A minor using exclusively white piano keys (Zhang and Song 2023). Translation also functions as a typographical device in Dark Souls 3, particularly in the transliteration of weapon names. Many medieval European weapons are imported into Japanese using katakana, the Multilingualism and Video Game Localisation 251 <?page no="252"?> character set primarily reserved for foreign words. Unlike hiragana and kanji, typically used for native concepts, katakana enhances the exoticism of the game. This deliberate linguistic strategy mirrors Japanese author Haruki Murakami’s use of katakana to highlight nonnormative speech (Walkowitz 2015: 15). By employing translation as an embedded literary tool, Miyazaki applies similar techniques in game design (Zhang & Song 2023). This borntranslated approach exemplifies how multilingualism is embedded into the very fabric of game design, shaping narrative, aesthetics, and player experience across linguistic and cultural boundaries. Multilingualism is also embedded in the development process when games are written in multiple languages simultaneously (Zhang and Song 2023). Signalis (roseengine, 2022) exemplifies this approach, having been crafted in both German and English at the same time, without a single "original" language. This practice is also evident in The Witcher series (CD Projekt Red 2007), where English and Polish versions were developed concurrently. Such an approach not only facilitates linguistic authenticity but also streamlines simultaneous shipment (simship), a common strategy for launching games in multiple markets at once. By integrating multilingualism into development, studios enhance linguistic and cultural resonance, ensuring that the game feels native to diverse audiences while maintaining consistency across translations from the outset. Lastly, game source texts are not necessarily final upon initial release; instead, they may be distributed in batches over time (Zhang and Song 2023). During localisation, translators may identify linguistic or cultural concerns, such as typographical errors or potentially offensive content, which are then commu‐ nicated to the development team for revision. Additionally, player feedback during beta testing may inform further refinements before the official release, leading to adjustments in both the source text and translations. Game texts often undergo multiple rounds of revisions, shaped by input from multilingual and multicultural teams. As digital interactions continue to expand, multilingualism has naturally integrated into video games, shaping both player communication and ingame collaboration. Within gaming environments, multilingual interactions involve navigating multiple languages, interpreting cultural nuances, and bridging linguistic barriers to foster engagement and cooperation among players from diverse backgrounds. Players have the autonomy to select the language in which they prefer to experience a game. Translated text intended for a specific locale may not always be used exclusively by players from that region, as many gamers opt to play in different languages based on preference or familiarity. Skilled players may even replay a game in multiple languages to deepen their 252 Xiaochun Zhang <?page no="253"?> immersion. Additionally, in online multiplayer settings, players often engage in one language while conversing with teammates in another, creating dynamic multilingual interactions. The digital world allows players to fluidly navigate linguistic spaces, forming flexible and diverse gaming communities unrestricted by geographic or linguistic boundaries. This adaptability reflects broader shifts in the understanding of multilingualism from focusing on multilingual writing to multilingual reading. This transformation is reshaping the field by recogni‐ sing the active role of the reader in interpreting multilingual texts (Yildiz 2024). Tidigs and Huss (2017) introduce the idea of “reader diversity,” challenging the assumption of a monolingual audience and positioning the reader as a “cocreator” of the text. Their approach embraces a multimodal perspective, considering the sensory and semiotic aspects of multilingual literature, in this case video games, such as its visual and auditory dimensions. Readers with diverse linguistic backgrounds interact with texts in unique ways, shaping their comprehension through pronunciation and interpretation. For video game players, this engagement extends beyond textual understanding to immersive interaction with the narrative universe woven into the game world. As a result, the act of reading itself produces a work that is uniquely mono or multilingual, reflecting the interplay between individual linguistic repertoires and textual composition. Multilingualism in video games is also rich resource for Natural Language Processing (NLP), as highlighted by Pöyhönen et al. (2022). By extracting per‐ suasive dialogue from role playing games across multiple languages, including English, Spanish, French, Italian, and German, the authors demonstrate how game text, often semiannotated by developers, provides structured linguistic data that can aid persuasion detection. The research underscores how multilin‐ gual game dialogues enable crosslinguistic comparisons, refine NLP models, and improve realworld applications. Furthermore, it advocates for interdisciplinary collaboration, showing that games can advance NLP while benefiting from enhanced language processing to create more immersive, dynamic player experiences. 3 Multilingualism and the practice of game localisation Employing over 330,000 people worldwide, the gaming industry plays a signi‐ ficant role in driving both economic growth and cultural influence (Uniglobal‐ union 2022). Video games have become one of the most popular forms of digital entertainment, with a global gaming population of 3.6 billion in 2025 (Newzoo 2025). Multilingualism in video games reflects the global nature of the gaming Multilingualism and Video Game Localisation 253 <?page no="254"?> industry, where players from diverse cultural and linguistic backgrounds come together. One of its most prominent features is localisation, the process of adapting video games for different linguistic and cultural audiences, ensuring they remain accessible and immersive worldwide. It involves more than simple translation, requiring adjustments in dialogue, humour, cultural references, and even visual elements. Effective localisation maintains the emotional and narrative integrity of a game while making it relevant to diverse players. This process enhances player immersion, allowing them to engage with characters and storylines in ways that feel natural in their own language (Mangiron and O’Hagan 2006). The game localisation sector typically encompasses five key professional roles, as outlined by BernalMerino (2014), including (1) managing, which involves positions such as associate producer, localisation coordinator, and production coordinator; (2) engineering, covering localisation engineers and sound engineers who ensure technical adaptation; (3) translating, consisting of text translators and translation editors responsible for linguistic accuracy; (4) testing, which includes QA coordinators, technicians, leads, and testers who verify quality and functionality. As previously noted, game localisation teams are inherently multilingual and multicultural, with language specialists contributing expertise from diverse linguistic backgrounds through collabora‐ tive efforts. Successful games are frequently localised, with multilingualism playing a crucial role in their global appeal. Dranch (2019) analysed 26,000 titles across Steam and the US App Store, revealing that 42 % support multiple languages. This percentage rises significantly among topperforming games, with localisa‐ tion present in 77 % of the most popular 1,000 titles. Among the top 100 most popular games, an impressive 96 % offer multilingual options, reinforcing the idea that linguistic accessibility is a key driver of success. Additionally, the most successful games tend to support a broader range of languages. Titles in the top 10 most popular games average 20 languages, while those ranked between 11 and 100 offer around 12. Games in the 101 to 1,000 range typically support eight languages. This pattern underscores the critical role of localisation in expanding reach and fostering engagement across linguistic and cultural communities, making gaming an increasingly global and inclusive medium. Multilingualism is negotiated through complex and diverse translation types, strategies and processes. In video games, there are typically two forms: dubbing and subtitling. Depending on the game’s target markets, game elements can be subtitled, dubbed, or both. While game subtitling shares some similarities with DVD subtitling, such as intralingual subtitles in dubbed versions and player‐ controlled options like pausing or skipping, its conventions differ significantly 254 Xiaochun Zhang <?page no="255"?> from those used in film and television. Mangiron and O’Hagan (2006) observe that game subtitles must keep pace with fastmoving gameplay, often appearing at a quicker rate than those in cinematic productions. Additionally, subtitles in games are constrained by pixelbased formatting rather than character count, necessitating strict space management. Mangiron (2013) found that standard subtitling guidelines from film and TV are not consistently applied to games, leading to subtitles of more than three lines, small font sizes, and segmenta‐ tion that disregards semantic coherence. These factors can negatively impact player experience. Dubbing practices in gaming also vary. Japanese games are often dubbed into English with subtitles in other European languages, while Englishlanguage games may be either dubbed or subtitled for international audiences (Mangiron and O’Hagan 2006). Despite the preferences of players in countries with strong dubbing traditions, subtitling remains the favoured choice for developers due to its lower cost and faster implementation. Multilingualism is manifestoed in the practice of pivot translation in game localisation. Pivot audiovisual translation refers to the process or outcome of translating audiovisual content through an intermediate language or text. As a form of indirect translation, it is often described as “a translation of a translation”, meaning the final translated text is not directly derived from the original source (Assis Rosa et al. 2017: 115). In the gaming industry, English frequently serves as a pivot language in video game localisation, particularly when adapting content originally developed in a language other than English (LOTE) (LocalizeDirect 2021: 15). Taking Dark Souls as an example again, the game has been localised into 11 languages through different translation pathways. English, Traditional Chinese, Simplified Chinese, and Korean were localised directly from the original Japanese text, maintaining a close linguistic connection to the source material. Meanwhile, French, Italian, German, Spanish (Spain), Polish, Russian, and Portuguese (Brazil) were translated from the English version, making English a pivot language in the localisation workflow. This layered translation process exemplifies the multilingual nature of game localisation, where languages interact across multiple stages. In terms of localisation strategies, multilingual elements are often conveyed through accents and dialects in dubbing, enriching the narrative’s authenticity. The Witcher 3: Wild Hunt exemplifies this approach by incorporating diverse linguistic variations to reflect its expansive world and cultural diversity. The English version of the game features a range of accents, including British dialects like West Country, American English, Scottish accents for Dwarves, and Northern Irish for the Skelligans (Wallis 2022). These distinct linguistic choices Multilingualism and Video Game Localisation 255 <?page no="256"?> help to reinforce character identities and regional distinctions within the game’s universe, making the world feel immersive and lived in. In game subtitling, multilingual elements are often either preserved to maintain an exotic feel or removed to ensure a smoother reading experience in the target language. The decision typically depends on the linguistic proximity between languages. For instance, in On Your Tail (Memorable Games 2024), several characters speak English mixed with their native languages, such as French. In the localised versions for European languages like Spanish, French words were retained to preserve the multilingual aspect, as many Spanish players can likely understand basic French terms like merci. Conversely, in the Asian localisations, such as Chinese, these multilingual elements were removed, likely to accommodate players who may be less familiar with European lang‐ uages. This approach demonstrates how subtitling choices shape multilingual representations in game localisation. Multilingualism poses challenges to the translators, as the translation process is influenced by both contextual and functional aspects of multilingualism. An author’s and a translator’s linguistic and cultural expertise are shaped by their personal and societal experiences, exposure to diverse cultures, and access to texts and languagelearning opportunities (Monticelli et al. 2023: 9). While translators must have a strong command of the primary language in a multilingual text, they and the audience may not fully grasp the additional languages or the cultural nuances embedded within them. This can lead to adjustments in the translation, often resulting in a simplification or misinterpre‐ tation of multilingual elements, diminishing their original impact. Translating multilingual cultural products reveals internal cultural tensions, which can sometimes spark conflict but also serve as catalysts for renewal and positive transformation. Multilingualism and video game localisation increasingly de‐ pend on advanced digital-humanities and language-technology infrastructures that enable collaborative, data-driven work with large-scale language resources (Budin 2014) to meet the growing demands of cross-cultural game development. 4 Conclusion Multilingualism has become a key focus in translation studies, reflecting a broader cultural shift toward linguistic diversity, acceptance, and curiosity about crosscultural interactions. It challenges traditional monolingual and national frameworks, emphasising the fluid exchange between languages and cultures in an interconnected world (Monticelli et al. 2023). This linguistic diversity is particularly evident in the gaming industry, where development 256 Xiaochun Zhang <?page no="257"?> teams collaborate across languages, fostering inclusivity in both the creation and localisation of video games. Multilingual elements appear across ingame dialogue, technical text, and marketing materials, often crafted by diverse teams whose native languages differ from those of the target audience. Many games are “born translated”, designed from the outset with international players in mind. Some titles are even developed in multiple languages simultaneously, reinforcing accessibility and cultural relevance. Multilingualism also influences player engagement, as gamers fluidly navigate different languages in digital environments. Players often choose preferred language settings, switch langua‐ ges during gameplay, or engage with multilingual text across subtitles and user interfaces. This interaction aligns with the broader academic shift from focusing solely on multilingual writing to considering multilingual reading as an active interpretative process. In this context, readers and players function as cocreators of meaning, shaping narratives based on their linguistic repertoires and cultural backgrounds. Translation plays a crucial role in enabling communication in multilingual societies, influencing language contact and facilitating linguistic evolution (Angermeyer 2023: 415). Game localisation is deeply intertwined with multi‐ lingualism, and has become an essential component of game development, transforming singlelanguage builds into culturally adapted versions for diverse audiences. This work requires interdisciplinary collaboration, as linguists, engineers, and testers refine text, audio, visuals, and mechanics to ensure linguistic and cultural cohesion. Pivot translation workflows further highlight localisation’s creative complexity. As games pass through multiple linguistic adaptations, each translation layer adds new interpretations and refinements. Localisation teams serve as mediators between the original design intent and varied player expectations, ensuring that gameplay feels culturally authentic yet universally engaging. Through these efforts, multilingualism in gaming fosters crosscultural dialogue and deepens global inclusivity, allowing players worldwide to experience games that feel both locally tailored and globally connected. This interplay between multilingualism, translation, and localisation reinforces the gaming industry’s role in shaping linguistic accessibility and cultural exchange in contemporary digital spaces. Multilingualism and Video Game Localisation 257 <?page no="258"?> Bibliography Assis Rosa, Alexandra/ Pieta, Helena/ Bueno Maia, Rita (2017). Theoretical, methodologi‐ cal and terminological issues regarding indirect translation: an overview. Translation Studies 10 (2), 113-132. Angermeyer, Philipp (2023). Multilingualism and translation. In: Aronin, Larissa/ Sing‐ leton, David (Hrsg.). The Routledge Handbook of Multilingualism, 415-427. New York/ London: Routledge. Aronin, Larissa (2019). What is multilingualism. Twelve Lectures in Multilingualism 1 (1), 3-34. Bernal-Merino, Miguel Á. (2014). Translation and localisation in video games: making entertainment software global. New York/ London: Routledge. Bernal-Merino, Miguel Á. (2009). Video games and children’s books in translation. The Journal of Specialised Translation, 11, 234-247. Budin, Gerhard (2014). Digital humanities, language industry, and multilingualism - global networking and innovation in collaborative methods. In: Schopp, Jürgen/ Pütz, Martin (Hrsg.). Pooling Academic Excellence with Entrepreneurship for New Part‐ nerships. Proceedings of the CIUTI Forum. Berlin: Frank & Timme, 45-62. Cenoz, Jasone (2013). Defining Multilingualism. Annual Review of Applied Linguistics, 33, 3-18. Council of Europe (2007). From Linguistic Diversity to Plurilingual Education: Guide for the Development of Language Education Policies in Europe. Strasbourg: Council of Europe. Council of Europe (2025). Language Facts. Available at: https: / / edl.ecml.at/ en/ Facts/ Lan guage-Facts (Accessed: 01.04.2026) Dranch, Konstantin (2019). Considering Video Game Languages. Multilingual. Available at: https: / / multilingual.com/ issues/ may-jun-2019/ considering-video-game-languages / (Accessed: 01.04.2026) Egenfeldt-Nielsen, Simon/ Smith, Jonas Heide/ Tosca, Susana Pajares (2019). Understan‐ ding Video Games: The Essential Introduction. London/ New York: Routledge. Gainsford, Peter (2019). Ancient Languages in Assassin’s Creed: Origins. Available at: h ttps: / / kiwihellenist.blogspot.com/ 2019/ 10/ ancient-languages-origins.html (Accessed: 01.04.2026) Gration, Elizabeth (2025). Bilingualism in 2025: US, UK & Global Statistics. Available at: https: / / preply.com/ en/ blog/ bilingualism-statistics (Accessed: 01.04.2026) Houška, Jan (2025). Language Management in Semi-Peripheral Game Production: How Foreign Workers in Czech Video Game Studios Experience the Use of English, Czech, and Other Slavic Languages. Multilingua, 44(4), 485-510. Jenkins, Laura (2022). Multilingual London. Available at: https: / / www.ihlondon.com/ blo g/ posts/ 2022/ multilingual-london (Accessed: 01.04.2026) 258 Xiaochun Zhang <?page no="259"?> Kirkpatrick, Andy (2011). English as a Medium of Instruction in Asian Education (from Primary to Tertiary): Implications for Local Languages and Local Scholarship. Applied Linguistics Review, 2, 99-120. Mangiron, Carme/ O’Hagan, Minako (2006). Game Localisation: Unleashing Imagination with a Restricted Translation. Journal of Specialised Translation, 6, 10-21. Mangiron, Carme (2013). Subtitling in Game Localisation: A Descriptive Study. Perspec‐ tives: Studies in Translation Theory and Practice, 21(1), 42-56. Meylaerts, Reine (2006). Heterolingualism in/ and Translation. How Legitimate Are the Other and His/ Her Language? An Introduction. Target, 18(1), 1-15. Meylaerts, Reine (2012). Multilingualism and Translation. In: Handbook of Translation Studies: Volume 1. Amsterdam/ Philadelphia: John Benjamins, 227-230. Meylaerts, Reine (2013). Multilingualism as a Challenge for Translation Studies. In: Routledge Handbook of Translation Studies. London/ New York: Routledge, 519-533. Meylaerts, Reine/ Serban, Adriana (2014). Introduction Multilingualism at the Cinema and on Stage: A Translation Perspective. Linguistica Antverpiensia-New Series, 13(1), 1-13. Monticelli, Daniele/ Saagpakk, Maris/ Verschik, Anna (2023). Literary Multilingualism in Translation: Texts, Contexts, Agents. Introduction/ Kirjanduslik Mitmekeelsus Tõlkes: Tekstid, Kontekstid, Methis. Studia Humaniora Estonica, 25(31-32). Mubeen, Asad (2025). 8 Video Game Characters Who Speak More Than One Language. Available at: https: / / www.thegamer.com/ best-multilingual-video-game-characters/ (Accessed: 01.04.2026) Newman, James (2004/ 2013). Videogames. 2. Aufl. London/ New York: Routledge. Newzoo (2025). Newzoo’s Global Games Market Report 2025. Available at: https: / / newzo o.com/ resources/ trend-reports/ newzoo-global-games-market-report-2025 (Accessed: 01.04.2026) O’Hagan, Minako/ Mangiron, Carmen (2013). Game Localization: Translation for the Global Digital Entertainment Industry. Amsterdam/ Philadelphia: John Benjamins. Olivares, Vladimir (2021). How a Japanese Myth Inspired Dark Souls’ Most Iconic Gameplay Mechanic. Available at: https: / / www.cbr.com/ dark-souls-weapons-tails-y amato-no-orochi/ (Accessed: 01.04.2026) Parkin, Simon (2015). Bloodborne Creator Hidetaka Miyazaki: ‛I Didn’t Have a Dream. I Wasn’t Ambitious’. The Guardian. Available at: https: / / www.theguardian.com/ te chnology/ 2015/ mar/ 31/ bloodborne-dark-souls-creator-hidetaka-miyazaki-interview (Accessed: 01.04.2026) Paterson, Eddie/ Williams, Timothy/ Cordner, Will (2019). Once Upon a Pixel: Storytelling and Worldbuilding in Video Games. Boca Raton: CRC Press. Pöyhörn, Teemu/ Hämäläinen, Mika/ Alnajjar, Khalid (2022). Multilingual Persuasion Detection: Video Games as an Invaluable Data Source for NLP. CoRR abs/ 2207.04453. Multilingualism and Video Game Localisation 259 <?page no="260"?> Rao, Parupalli Srinivas (2019). The Role of English as a Global Language. Research Journal of English, 4(1), 65-79. Ruggill, Judd/ McAllister, Ken/ Nichols, Randy/ Kaufman, Ryan (2016). Inside the Video Game Industry: Game Developers Talk About the Business of Play. Abingdon: Routledge. Taylor, Mark (2022). UK Games Industry Census: Understanding Diversity in the UK Games Industry Workforce. London: UKIE. Tidigs, Julia/ Huss, Markus (2017). The Noise of Multilingualism: Reader Diversity, Linguistic Borders and Literary Multimodality. Critical Multilingualism Studies, 5(1), 208-235. Walkowitz, Rebecca L. (2015). Born Translated: The Contemporary Novel in an Age of World Literature. New York: Columbia University Press. Wallis, Adam (2022). The Localization of the Witcher: Accents and Cultural Connections. Available at: https: / / terralocalizations.com/ 2022/ 12/ 13/ the-localization-of-the-witche r-accents -and-cultural-connections (Accessed: 01.04.2026) Ubisoft (2025). Diversity, Inclusion and Accessibility: Celebrating Diversity and Putting Belonging at the Heart of What We Do. Available at: https: / / www.ubisoft.com/ en-u s/ company/ about-us/ diversity-inclusion-accessibility (Accessed: 01.04.2026) Uni Global Union (2022). The Video Game Industry: A Resource for Organizers. N.p.: Uni Global Union. Available at: https: / / uniglobalunion.org/ report/ the-video-game-indust ry/ (Accessed: 01.04.2026) United Nations (2025). International Migration. Available at: https: / / www.un.org/ en/ glo bal-issues/ migration (Accessed: 01.04.2026). Yildiz, Yasemin (2024). Special Focus Introduction: Translating Multilingualism. Studies in 20th & 21st Century Literature, 48(1), 4. Zhang, Xiaochun/ Song, Huawei (2023). The Mukokuseki Strategy and the Application of Pivot Translation in the Localization of Japanese Games. Games and Culture, 10 (6). DOI: 10.1177/ 15554120231216545 Ludography Assassin’s Creed: Revelations (Ubisoft 2011) Assassin’s Creed: Syndicate (Ubisoft 2015) Bloodborne (FromSoftware, 2015) Dark Souls 3 (FromSoftware, 2016) Final Fantasy X (Square Enix, 2001) Hitman (IO Interactive 2021-present) Shadow of the Tomb Raider (Eidos-Montréal, Nixxes Software, 2018) Signalis (rose-engine, 2022) 260 Xiaochun Zhang <?page no="261"?> The Witcher (CD Projekt Red, 2007) Tomb Raider (Embracer Group, 1996-2025) On Your Tail (Memorable Games, 2024) Multilingualism and Video Game Localisation 261 <?page no="263"?> Contributors Elena Chiocchetti, Institut für Angewandte Sprachforschung, Eurac Research Dragoș Ciobanu, Human and Artificial Intelligence in Translation (HAITrans), University of Vienna Rute Costa, NOVA University Lisbon Juan Carlos Díaz-Vásquez, EAFIT University Wolfgang U. Dressler, Austrian Centre for Digital Humanities der Österreichi‐ schen Akademie der Wissenschaften Jan Engberg, School of Communication and Culture, Aarhus Universität Hendrik Kockaert, Faculty of Arts, KU Leuven Waltraud Kolb, Center for Translation Studies, University of Vienna Georg Löckinger, University of Applied Sciences Upper Austria Maria Cecilia Plested Alvarez, Humboldt University of Berlin Natascia Ralli, Institut für Angewandte Sprachforschung, Eurac Research Peter Sandrini, Universität Innsbruck Larisa Schippel, Weave-Projekt Graz: Post-Exile Trans Klaus-Dirk Schmitz, Technische Hochschule Köln Frieda Steurs, KU Leuven, Instituut voor de Nederlandse Taal (INT) Werner Winiwarter, Faculty of Computer Science, University of Vienna Tanja Wissik, Austrian Academy of Sciences Xiaochun Zhang, University College London <?page no="264"?> ISBN 978-3-381-15371-8 Gerhard Budin is a leading authority in terminology science, translation studies, digital humanities, and the philosophy of science. His influential work has shaped research agendas, advanced international standards, and fostered interdisciplinary collaboration. This volume celebrates his outstanding contributions and honors his 65 th birthday by bringing together scholars from terminology science, translation studies, and (computational) linguistics. Budin has played a key role in carrying the rich Viennese tradition of terminology science into the modern era through critical reflection and innovation. To honor this legacy, the volume includes original chapters that revisit fundamental research questions and showcase terminology work. His impact on the European language technology landscape is reflected in contributions on language representation and technology, while his linguistic background informs chapters on intertextuality and related topics. Translation forms a third major pillar, honored with original research on translation competence, technology, history, and localization. Gromann / Heinisch / Lušicky (Hrsg.) Beyond Terminology Dagmar Gromann / Barbara Heinisch / Vesna Lušicky (Hrsg.) Beyond Terminology A tribute to Gerhard Budin