Interview with Christophe Roche

Christophe RocheChristophe Roche: University of Savoie (France)

roche@univ-savoie.fr, www.condillac.org.

Christophe Roche is Full Professor at the University of Savoie, France, and associate researcher at the Linguistic Research Center of the New University of Lisbon, Portugal. He is in charge of the Condillac Research Group on “Ontology and Terminology”. His main domains of interest are knowledge engineering and ontology, terminology and linguistics, and their applications. Involved in several European projects, Christophe Roche is the Project leader of the ISO 704 and ISO 1087 Standards on Terminology, its principles and methods. Since 2007 he chairs the international scientific committee of the TOTh Conferences (Terminology & Ontology: Theories and applications).

 

1. What is your background and why are you interested in terminology?

My background is in artificial intelligence (AI), specifically the area of knowledge representation – the extraction, modelling and formalisation of knowledge in a form that a computer system can understand. Working with concepts, their definition and naming, in particular in the context of scientific and technical discourse, inevitably means working in terminology.

2. Why do ontologies play an important role in terminology?

There is no term that is not tied a concept. This means that terminology relies on two systems, a linguistic system and a conceptual one. The AI field of ontology, developed in the context of knowledge engineering and defined as a formal definition of a conceptualisation, offers one of the most promising ways forward when it comes to representing conceptual systems. By providing a means for formulating explicit and computational representations of concepts, ontology can serve as a tool not only for operationalising terminology for IT applications, but also for focusing on concepts independently of language.

3. Can you define ‘ontoterminology’ as described in your paper ‘Howto unify terminology and ontology into a single paradigm’?

An ontoterminology is a terminology whose conceptual system is a formal ontology. Using ontologies it is possible to distinguish between, and to link without confusion, the linguistic and conceptual systems on which any terminology is based. We need to bear in mind that the network of terms of a given language is not isomorphic to the network of concepts, which is assumed to be common and shared by different communities: discourses about the ‘world’ are distinct from scientific conceptualisation of the ‘world’.

Ontoterminology requires a ‘rethink’ of terminology, its principles and methods. For example, the ontological approach allows the terminologist to distinguish between terms and concept names – since they belong to two different semiotic systems – as well as between the definition of a term and the definition of a concept. The definition of a term, written in a natural language, is either a linguistic explanation of what the concept is (‘thing definition’) or a definition of the meaning of the term in discourse (‘word definition’), whereas a concept definition is the specification of a concept written in a formal language.

4. Could you explain briefly what the multilingual information retrieval system in the domain of renewable energy carried out as part of the ASTECH project[1] has to do with ontoterminology?

The main objective of the ASTECH project was to set up a multilingual content management system on renewable energy. Given that the domain ontology is common and shared by all the communities of practice, which differ only in the terms used for designating concepts, it is possible to classify documents on language-independent concepts based on the terms that appear in the documents.

It is then possible to conduct a search in any of the languages used in the retrieval system to find all documents relevant to a specific subject, regardless of which language the text is in, insofar as they are classified under the same concepts. Furthermore, the recall and precision criteria for the retrieved documents are improved as a consequence of the logical properties of the ontology.

5. Can you explain the difference between the classical semantic triangle (Ogden and Richards, 1923) and the double semantic triangle, and the relevance this has for terminology?

The semantic triangle of Ogden and Richards ‘Symbol – Thought – Referent’, like the more classical semantic triad ‘vox – conceptus – res’ of scholastic philosophy, serves to describe a single semiotic system, clarifying the relationships between a sign and what it refers to. As terminology involves two different and non-isomorphic semiotic systems, however, we need a double semantic triangle in order to express the notions involved and how they relate to each other. In a double semantic triangle a concept is not a signified (term meaning), and the identifier of a concept is not a signifier (term), just as the linguistic relationships between terms are not linguistic translations of conceptual relationships (for example, hyperonymy between terms cannot be reduced to a linguistic translation of subsumption between concepts).

6. How would you see an attempt to ‘ontologise’ IATE?

The ontologisation of IATE would mean building ontologies for each domain, placing focus more on concepts, their definition in a formal language and their logical relationships, rather than on the words designating them. Such an approach would require contributions from experts in each domain. To avoid the pitfalls that natural language presents, it would also require dedicated tools that rely on formal languages and systems. In discourse, people too often equate concepts with the terms that designate them, when in fact concepts are by nature extra-linguistic.

7. Are you familiar with the terminology activities of the EU institutions and the presence of EP-TermCoord?

Not enough, I’m sorry to say. I would be happy to learn more.

[1] ‘Advanced sustainable energy technologies for cooling and heating applications’, funded under FP6-SUSTDEV. See European Commission, CORDIS: http://cordis.europa.eu/project/rcn/85628_en.html.


Interviewer: Soraia MonteiroSoraia Monteiro

Born in Lisbon and graduated in Translation at the New University of Lisbon (languages of work: English, French and Spanish); currently doing a M.A. in Terminology and Specialized Information, in order to strengthen the terminology field and that is why she wanted to work at TermCoord in the first place. Soraia has already worked as an in-house translator in several translation companies, as a voluntary translator in United Nations and for TED Talks. Concerning the future, her goal is to work for international organisations as a terminologist or a translator.