The first week of May will be very busy at the OLST

Collocations dictionaries are essential tools to undertake specialized translation and scientific and technical writing. In the ONTODIC II Project we aim to develop a systematic methodology for the elaboration of terminological collocations dictionaries, that is, dictionaries that contain specialized terminology in a knowledge area and which allow the user to access directly to a specific combinatorial structure of a term.
The design of the methodology and the techniques for the elaboration of collocations dictionaries include the following aspects: 1) analysis of linguistic elements in the entries, 2) formalization of these elements in valid structures for any group of terms, 3) implementation of these elements and structures in ontologies so that it proves useful for the creation and design of other dictionaries and the access to data through natural language processing tools.
We follow the Explanatory and Combinatorial Lexicology for the analysis and linguistic formalization of collocations due to the high level of formalization of this theory. This allows a good approach to the implementation of the elements which will be included in the dictionary entries. In order to implement data in an ontology, we use the ontology editor Protégé.

The meanings of lexical units (LUs) are constructed in relation to background knowledge. Frame Semantics is based on the assumption that the structure of this background knowledge can be analyzed in terms of semantic frames. A frame is a schematic representation of a situation that includes its participants, props, and other conceptual elements. Each LU evokes a particular frame and profiles some element or aspect of that frame. The Spanish FrameNet Project (SFN) has been developing an online lexical resource for Spanish that is based on frame semantics and supported by corpus evidence. SFN has also recently begun creating a similar resource for grammatical constructions, describing their grammatical characteristics and semantic import and linking them to frames. The project database is online and can be queried at http://sfn.uab.es:9080/SFN.

Le projet Base de Données Terminologique et Textuelle (BDTT-AR) se subdivise en deux phases différentes, la première, d’une durée de 2 ans, s’est déroulée de 2005 à 2007 et la deuxième, d’une durée de 3 ans, se trouve actuellement en phase de développement, et ce depuis le début du second semestre 2009 et jusqu’en septembre 2012. À chacune de ces deux périodes distinctes du projet correspondent une conjoncture et des objectifs bien spécifiques, autant pour l’institution qui accueille le projet, l’Assemblée de la République (http://www.parlamento.pt), que pour le Centre de Linguistique de l’Universidade Nova de Lisboa (CLUNL, http://www.clunl.edu.pt/PT/home.asp), institution qui le propose.
Ce projet est un exemple de coopération entre le secteur de l’administration publique et celui de la recherche universitaire avec, d’une part, le CLUNL ayant des objectifs axés sur la recherche, tant sur le plan théorique que méthodologique, et ancrés dans la formation en Terminologie pour l’administration publique, mais également dans le développement d’une ressource terminologique adaptée à la réalité et aux besoins de l’institution et devant prévoir, à la longue, aussi bien la question de la gestion de la base de données en contenus terminologiques et textuels que leur processus de validation. Et, de l’autre, l’institution qui accueille le projet, l’Assemblée de la République, pour qui la décision de créer sa propre base de données terminologique et textuelle constituait déjà en soi un défi important, mais encore moindre que celui de mettre en place une méthodologie de travail intégrant un processus de garantie de la qualité des contenus linguistiques, terminologiques et textuels de la BDTT-AR et, le tout, comme support au processus de traduction.
C’est cette méthodologie de travail que nous irons aborder lors de notre intervention.
** Ce travail de recherche est mené conjointement avec Raquel Silva (CLUNL) en étroite collaboration avec Zara Almeida (CLUNL & AR)

Multilingual resources covering the subject field of law do not always include specialized verbs and they do not always include valid equivalents. However, legal terminology, namely the terminology that can be found in Supreme Court judgments, not only consists of noun terms but also of specialized verbs. We propose a methodology based on the theory of Frame Semantics (Fillmore 1977, 1982, 1985) and its application FrameNet (Ruppenhofer et al. 2010) to describe that kind of predicative units and to assign their (Portuguese-English-Portuguese) equivalents. In this research, every 200 specialized verbs that were selected are said to evoke a frame, a sort of conceptual scenario in which a certain number of mandatory participants play a given role (the role of the judge, the role of the appellant, the role of the law). The labels that were attributed to the 76 frames that we identified allowed us to group together specialized verbs within each language as well as 165 pairs of equivalents. 71% of the equivalent pairs were considered total equivalents because they evoke the same frame in the same way, their actantial structures are identical, the linguistic realizations of the actants are equivalent, and the syntactic patterns of the verbs are similar. 29% of the equivalent pairs were considered partial equivalents because at least one of these criteria was not met. We present the equivalence scenarios that we observed by means of examples and show in which ways future users will be able to use JuriDico, the lexical resource that was compiled for this project.