Identifying Equivalents of Specialized Verbs in a Bilingual Comparable Corpus of Judgments: A Frame-based Methodology

Janine Pimentel
Observatoire de linguistique Sens-Texte Centro de Lingustica da Universidade Nova de Lisboa janine.pimentel@umontreal.ca Abstract
Multilingual terminological resources do not always include the equivalents of specialized verbs that occur in legal texts. This study aims to bridge that gap by proposing a methodology to assign the equivalents of this kind of predicative units. We use a comparable corpus of judgments produced by the Supreme Court of Canada and by the Supremo Tribunal de Justia de Portugal. From this corpus, 200 English and Portuguese verbs are selected. The description of the verbs is based on the theory of Frame Semantics (Fillmore 1977, 1977, 1982, 1985) as well as on the FrameNet methodology (Ruppenhofer et al. 2010). Specialized verbs are said to evoke a semantic frame, a sort of conceptual scenario in which a number of mandatory elements play specific roles (e.g. the role of judge, the role of defendant). Given that semantic frames are language independent to a fair degree (Boas 2005; Baker 2009), the labels attributed to each of the 76 identified frames (e.g. [Crime], [Regulations]) were used to group together 165 pairs of candidate equivalents. 71% of them are full equivalents, whereas 29% are only partial equivalents. Keywords: equivalence, specialized verbs, Frame Semantics, comparable corpus

1.

Objectives

Multilingual resources that describe legal terminology do not always include all the equivalents that translators need to produce translations of legal texts. There are two main reasons for this. Firstly, legal systems are known to differ from one linguistic community to another and even from one country to another because each has its own history and traditions. As a result, legal terminologies are often anisomorphic, i.e. the conceptual structures may vary from one national legal system to another. Secondly, by focusing especially on the definition of equivalence, a notion widely discussed in translation studies but insufficiently debated in terminology, the literature does not provide sound and systematic methodologies to identify equivalents. There is thus a lack of criteria to search for and validate equivalent terms that occur in specialized texts. This problem is even more evident in the case of predicative units, such as verbs. Although some terminologists (LHomme 1998, Lorente and Bevilacqua 2000, Costa and Silva 2004) have worked on specialized verbs, terminological equivalence regarding this kind of units would benefit from a thorough study. In this paper, we propose a methodology for describing specialized verbs and for identifying their equivalents. The methodology is based on the theory of Frame Semantics (Fillmore 1977, 1982, 1985; Fillmore and Atkins 1992) (hereafter FS) as well as on the FrameNet project (Ruppenhofer et al. 2010). Our hypothesis is that frames, i.e. conceptual scenarios in which words participate, can function as interlingual representations (Boas 2005; Baker 2009) grouping together not only synonym and near synonym terms but also equivalents. The ultimate goal of the research is to present criteria for validating equivalents of this kind of units, so as to improve multilingual terminological resources, in general, and lexical resources covering the subject field of law, in particular.

This paper is organized as follows. Section 2 defines what we understand by terminological equivalence, briefly accounts for the criteria guiding the identification of equivalents that have been put forth by lexicographers and terminologists, and explains how FS can be of assistance to our project. Section 3 describes the methodology, namely the constitution of the corpus, the selection of specialized verbs, the description of the verbs and the identification of their equivalents. Section 4 presents the results of the research obtained up to now. Finally, section 5 draws some concluding remarks and discusses future research perspectives.

2.

In search of terminological equivalence

As Adamska-Saaciak (2010: 387) states to be able to talk about equivalence, there must be (at least) two entities of some kind, a certain relationship between those entities, and a certain value of that relationship. In our research, the entities between which the relationship of terminological equivalence obtains are specialized verbs, i.e. lexical units (hereafter LUs) that are very relevant in the discourse of judges. This perspective is closer to that adopted by lexicographers, who view the entities in question as LUs (Wiegand 2005), than to that adopted by classical terminologists (e.g. Picht and Draskau 1985) who consider the entities between which the relationship of terminological equivalence obtains to be concepts. Both lexicographers and terminologists agree that the relationship between these entities is one of more or less symmetrical correspondence. Although the tripartite division of equivalence degrees (full, partial and zero equivalence) is accepted by most, the literature is rather silent when it comes to the presentation of methodologies for identifying equivalents. In addition, the few ideas that lexicographers and terminologists propose as to what should be taken into account in the task of identifying equivalents can differ quite considerably. For instance, Piotrowski (1994) and

1791

Atkins and Rundell (2008) consider it relevant to examine the discursive and formal dimension of the LUs, i.e. their collocational context, their value and function in discourse. Piotrowski (1994) suggests that equivalence should be equated as a third external entity or tertium comparationis. Atkins and Rundell (2008) argue for the use of corpora for finding equivalents. In contrast, authors like Sandrini (1995, 1996, 1999) and Chrom (2004) reject the existence of full equivalents between terms belonging to different legal systems and prefer to document national concepts for each language separately, instead of providing translators with equivalents. Although we agree that the documentation of concepts is unavoidable in legal terminography, it seems to us that a terminological resource built in these moulds is not entirely helpful for translators. As they often have to meet short deadlines to produce translations, translators will not have the time to study the documentation, make a decision on the best equivalent and then look up another resource to find the correct usage of the equivalent term. Therefore, we believe that a suitable resource for legal translators is a lexical resource that balances the documentation of concepts with the description of the linguistic behavior of terms. FS can be of assistance in the elaboration of such a resource as it offers a formal basis for linking the linguistic functioning of terms to conceptual scenarios. FS and its application FrameNet (Ruppenhofer et al. 2010) have been adapted in several terminological projects (e.g. Dolbey et al. 2006, Faber et al. 2006, Schmidt 2009). Some of these contributions include a multilingual dimension. For instance, the Kicktionary (Schmidt 2009), a resource describing football terminology, illustrates how valuable the FrameNet approach is to account for interlinguistic correspondences (and problems of true correspondence) between items pertaining to specialized subject fields.

for the benefit of legal profession, for the benefit of other judges and for the benefit of appellate courts. The corpus used in the research is composed of a single genre of texts, so as to facilitate the interpretation of the terminology therein (cf. Condamines 2008, Costa 2004, Rogers 2000). As text genres are social products, their instantiations can vary from one community of experts to another. For instance, although the Canadian and Portuguese judgments have a similar macrostructure, the Canadian judgment is in average twice longer than the Portuguese judgment. This can be explained by the fact that Canadian judges must look up not only the published legislation but also former judgments that relate to the case on which they are working and justify their selection. In contrast, this principle of binding precedent or stare decisis is not a formal principle that Portuguese judges must follow.

3.2 Selection of specialized verbs

Candidate-terms were extracted from the comparable corpus by means of a term-extractor called TermoStat (Drouin 2003). They were then validated by applying the lexico-semantic criteria provided in LHomme (2004), according to which a given lexical item may be a term if: 1) it has a meaning related to the subject field in question; 2) its actants (or arguments) are terms themselves according to criterion 1; 3) its morphological derivates are terms themselves according to criteria 1 and 2 at the same time that there is a semantic relation between the lexical item and its derivatives; and 4) the lexical item has other paradigmatic relations to other terms validated by the same criteria. The characteristics of judgments as a text genre are particularly helpful for applying criterion 1) as they provide clues about what happens in judgments (for an illustration of the application of these criteria please see Pimentel, forthcoming). Therefore, not only terms such as to acquit and to convict were selected but also other verbs, such as to adduce, to argue, to satisfy that are relevant in the corpus given the importance (and amount) of argumentation in the judgments.

3.

Methodology

In this section we present a methodology to describe specialized verbs and to assign their equivalents. It is based on the theory of FS (Fillmore 1977, 1982, 1985; Fillmore and Atkins 1992) as well as on the methodology developed by FrameNet lexicographers (Ruppenhofer et al. 2010). Some of the adaptations that we made to our project are mentioned throughout the following sections.

3.3 Frame-based description of the verbs

FS (Fillmore 1977, 1982, 1985, Fillmore and Atkins 1992) is based on the principle according to which the linguistic functioning of words can be linked to conceptual scenarios or frames. LUs are defined against the frames that they evoke which correspond to background knowledge. For instance, defining Tuesday (the LU) presupposes defining the notion of cyclic calendar (the frame). In addition, as extralinguistic entities, frames are considered to be language-independent to a fair degree (Boas 2005; Baker 2009). Frames contain mandatory meaning slots or core Frame Elements (hereafter FEs) as well as optional slots or non-core FEs. We adopt this principle and adapt the methodology developed by the FrameNet lexicographers (Ruppenhofer et al. 2010). Whereas FrameNet lexicographers describe a frame and then the LUs that evoke it, we describe the selected verbs and then group them into frames. Therefore, in this research, verbs were

3.1 Corpus design

Whereas FrameNet mainly uses the BNC corpus (British National Corpus), we use a corpus of authentic judgments produced by the Supreme Court of Canada and by the Supremo Tribunal de Justia de Portugal. This comparable corpus totals approximately 2,500,000 words for each language (i.e. English and Portuguese). Judgments are final decisions in legal disputes which are argued and settled in a court of law and determine winners and losers (Songer 2008: 78). They are written not only for the benefit of the parties involved in the case, but also

1792

first analyzed by examining their behaviour in concordances. Then, their actantial structures (or number of arguments) were identified and each actant (or argument) was attributed a semantic label (e.g. Arguer, Evidence, Law). These elements were assumed to be the core-FEs of the frame evoked by the verb. For instance, the verb to codify has two actants: Law and Issue. The labels attempt to describe the classes of entities to which the linguistic realizations of the actants refer (Table 1). For each verb we selected about twenty contexts illustrating its specialized meaning, its cooccurrents, other related terms, etc. We encoded the data in an XML editor and annotated the actantial structures of the verbs, i.e. the core-FEs, as well as optional information, such as the criteria used by the judge to make a decision (statutes, codes, previous decisions), i.e. the non-core FEs. We attributed labels to the frames that the verbs evoked and provided a tentative definition of them. Only then were the verbs grouped together into frames (Table 1). We did this for each language separately.
Frame Definition Core FEs Regulations Written legal documents (Law) constitute or constrain a certain Issue. Law Written legal documents that have the force of law (e.g. Constitution, jurisprudence) or will have the force of law once enacted (e.g. bill, legislature). They arrange and regulate a set of Issues. Issue It refers to what is required or forbidden by Law but also to rights, responsibilities and principles that are defined in the Law. codify1 (Law ~ Issue) establish1 (Law ~ Issue) govern1 (Law ~ Issue) mandate1 (Law ~ Issue) prescribe1 (Law ~ Issue) provide1 (Law ~ Issue) recognize1 (Law ~ Issue)

computational linguist, Benot Robichaud, who works in the Observatoire de linguistique Sens-texte (OLST), to program search queries that group together those terms that contain the same frame label in their descriptions. For instance, the Portuguese verbs consignar1, determinar1, disciplinar1, estabelecer1, estatuir1, estipular1 and prescrever1 had all been attributed the same frame label, i.e. [Regulations], which is also the label attributed to codify1, establish1, govern1, prescribe1, provide1 and recognize1. This means that the [Regulations] frame groups together 14 specialized verbs, i.e. 7 English verbs and 7 Portuguese verbs (see Appendix). Thus, we are dealing here with 49 pairs of candidate equivalents. As each of the aforementioned verbs has 7 candidate equivalents, the pairs of equivalents were studied closely because it was necessary to examine if they were full equivalents. If they were not full equivalents it was necessary to understand what differentiated the verbs in the pair. In order to do so, we compared the terms in the two languages by taking into account the following criteria:

1. The way the verbs evoke the frame. Here, we

examined if the verbs evoked the frame in the same way. For instance, if the verbs in one language corresponded to opposites of verbs in the other language, then this meant that they profiled the frame in a negative or positive way. In this case, only candidate equivalents that profiled one of these possibilities were retained. For example, the verbs grouped together in the [Regulations] frame (see Appendix), do not display any difference in the way they evoked the frame because they are not opposites. 2. The actantial structures of the verbs. After selecting the relevant candidate equivalents, we proceeded to examine their actantial structures, namely the number and nature of the actants. For instance, all aforementioned verbs have the same actantial structures (see Appendix). Even if the application of this criterion did not reveal any difference between the verbs, we continued to examine them in more detail by means of criteria 3 and 4. 3. The linguistic realizations of the FEs. The linguistic realizations of the actants that correspond, by implication, to the core FEs were analyzed, so as to confirm or refute whether the nature of the verbs actants is entirely identical. For instance, the linguistic realizations of the actants of the verbs grouped together into the [Regulations] frame all refer to the same kind of entities (see Appendix). 4. The syntactic patterns of the verbs. Finally, it was necessary to verify if there was any significant difference between the syntactic patterns of the verbs. Major differences between the syntactic patterns of the verbs, such as the lack of coincidence of the syntactic patterns of two or more actants of a verb, can raise translation challenges and may indicate that we are not dealing with full equivalents. For instance, no significant differences were identified in the syntactic patterns of the verbs grouped together into the [Regulations] frame.

Terms

Table 1: The frame [Regulations] groups together 7 English verbs Frames can group together certain synonyms, near-synonyms, opposites and related LUs. So, codify1, establish1, govern1, prescribe1, mandate1, provide1 and recognize1 were all grouped into the same frame because they have exactly the same core FEs: Law and Issue. We decided to call this frame Regulations and wrote a definition of it by taking into account the roles played by the core FEs as well as their linguistic realizations in the contexts (Table 1).

3.4 Identification of equivalents

Once the verbs in each language were grouped into frames, we then used the frame labels to link the verbs in the two languages. Thus, verbs with the same frame label were taken to be candidate-equivalents, this reflecting the idea that frames can function as interlingual representations (Boas 2005; Baker 2009). In order to accomplish this task, we asked the

1793

Whenever all these criteria were met, the verbs were considered full equivalents. However, if there were significant differences in any of these elements, the verbs were considered partial equivalents.

4.

Results

The methodology described in the previous section allowed us to group together 200 verbs into 76 frames. 165 pairs of equivalents were identified. Among these, 117 pairs of equivalents (71%) not only evoked the same conceptual scenario but the actantial structures of the verbs, the linguistic realizations of the actants and their syntactic patterns were similar. Therefore, they were considered full equivalents. For instance, the verbs included in the Appendix section are full equivalents because no particular differences in the elements mentioned in section 3.4 were identified. However, 48 pairs of equivalents (29%) did not entirely meet these criteria and were considered partial equivalents. More precisely, these verbs did not have the same number of actants; the semantic nature of the actants differed; and/or their syntactic patterns were considerably different. About 8% of the total number of equivalent pairs (i.e. 13 out of the 165 pairs of equivalents) were given the partial equivalence status because the comparison of the actantial structures of the verbs revealed that some verbs did not
Frame:

have the same number of actants. For instance, remit1 and transitar2 evoke the same conceptual scenario in which a court transfers a case or a matter to another court for further consideration or decision. However, the court from which a case or a matter is transferred (COURT1) is never instantiated in the Portuguese corpus, although it is conceptually mandatory because a case or a matter cannot be transferred to a different court (COURT2) if it wasnt already in a previous one (COURT1). As the entity responsible for the action of transfer is never expressed, the Portuguese verb is intransitive whereas the English verb is transitive. As a result, these verbs may raise translation difficulties because when translating transitar2 into English, the translator will necessarily have to use a term that denotes COURT1 because COURT1 is one of the actants of the equivalent remit1. However, as COURT1 is never expressed in the Portuguese contexts, the translator will not know which term they should use. About 16% of the total number of equivalent pairs (i.e. 27 out of the 165 pairs of equivalents) were given the partial equivalence status because the comparison of the actantial structures of the verbs revealed that the nature of the actants differed from one verb to another. For instance, commit1, engage1, incorrer1 and praticar1 were grouped together into the frame [Crime] (Table 2).

Crime Definition: A PROTAGONIST does something that is prohibited by law (CRIME). Core FEs: PROTAGONIST The PROTAGONIST can be anyone: the accused or the accuser (criminal suit), the petitioner or the respondent (civil suit) and even the judge. CRIME Conduct or behaviour that is prohibited by law. Non-core FEs: CRITERION, MANNER, MEANS, PLACE, REASON, TIME Contexts: It did not appear that a stranger could have committed the murder. H.L. became addicted to alcohol, had emotional difficulties, and engaged in criminal conduct. Por o mais, ficou sobejamente provado que o arguido incorreu na prtica dos crimes pelos quais foi condenado - todos no espao de cerca de 1 ms e meio, mais precisamente entre 011/05/2008 e 19/06/2008 -, nica e exclusivamente para conseguir custear o consumo de cocana e herona. Os arguidos no praticaram os crimes pelos quais foram condenados, onde no haver, como no h, prejuzo sofrido pela assistente nem timo fundante. PROTAGONIST accused, appellant, defendant, firm, individual, mr. dinardo, mr. last, person, police chief, police officer, respondent, suspect, stranger, witness Subject, NP accused, City, Crown, euro-excellence, firm, H.L., one, Sanofi, they Subject, NP arguido, autor, condutor, devedor, lesante, ru Subject, NP agente, arguido, autor, devedor, filho, lesante, pessoa, recorrente, relao, ru Subject, NP Terms CRIME abuse, act, assault, breach, break and enter, crime, fraud, homicide, fault, offence, manslaughter, murder, neglect, perjury, robbery, theft, tort Object, NP activity, behaviour, bid shopping, conduct, dispute, infringement, misconduct, patenting, practice, production Complement, PP abuso, crime, desobedincia, falta, inadimplemento, incumprimento, infraco, leso, mora, prtica, responsabilidade, violao Complement, PP acto, crime, facto, ilcito, infraco, violao Object, NP

English

commit1

engage1

Portuguese

incorrer1

praticar1

Table 2: The [Crime] frame

1794

The PROTAGONIST can be anyone: the accused or the accuser (criminal suit), the petitioner or the respondent (civil suit) and even the judge. The CRIME is the conduct or behaviour that is prohibited by law and is typically instantiated by terms denoting the generic idea of crime (e.g. crime, misconduct, infringement) or by terms denoting specific types of crime (e.g. break and enter, theft, homicide). We observed that the instantiations of the FE CRIME in the contexts in which the term commit1 occur denote specific types of crime, whereas the instantiations of the FE CRIME in the contexts in which the terms engage1, incorrer1 and praticar1 occur denote the generic idea of crime. As a result, the term engage1 can be translated by incorrer1 as well as by praticar1, but commit1 cannot be translated by incorrer1 or by praticar1 in all contexts. Finally, about 6% of the total number of equivalent pairs (i.e. 10 out of the 165 pairs of equivalents) were given the partial equivalence status because the comparison of the actantial structures of the verbs revealed that the verbs syntactic patterns were different. For instance, commit2, impose1, order1, require1 and determinar2 were grouped together into the frame [Order]. This frame describes a scenario in which the LAW or its representative, the JUDGE, commands somebody to do something. Thus, this frame involves four core FEs: JUDGE, LAW, DUTY and PROTAGONIST. The full equivalent of determinar2 is order1, whereas commit2, impose1 and require1 were considered partial equivalents because the observation of these verbs revealed significant differences in their syntactic patterns when compared to the syntactic patterns of determinar2. The Portuguese verb admits three syntactic patterns: JUDGE (Sub. NP) PROTAGONIST (Obj. NP) DUTY (Comp. That-clause) JUDGE (Sub. NP) DUTY (Comp. That-clause) JUDGE (Sub. NP) DUTY (Obj. NP) The first pattern occurs only once, whereas the second and the third pattern occur very often. Similarly, the English term order1 admits the first pattern three times, but the most frequent patterns are the second and the third. In contrast, commit2 only admits the pattern JUDGE (Sub. NP) PROTAGONIST (Obj. NP) DUTY (Comp. PP) which is similar to the first mentioned pattern of determinar2 but only occurs once in its contexts. The term impose1 only admits the patterns JUDGE (Sub. NP) DUTY (Obj. NP) PROTAGONIST (Comp. PP) and LAW (Sub. NP) DUTY (Obj. NP) PROTAGONIST (Comp. PP) which never occur in the contexts of determinar2. Among the syntactic patterns that the term require1 admits the only one that coincides with those of determinar2 is JUDGE (Sub. NP) DUTY (Obj. NP) but this pattern occurs only twice in the contexts of require1, whereas it is very frequent in the contexts of determinar2. For the moment, we cannot draw considerations on zero equivalents because only 200 specialized verbs were analyzed and because the purpose of the research was to find valid equivalents rather than zero equivalents. Nevertheless, we make the hypothesis that when none of the aforementioned criteria are met, then we are dealing with zero equivalents.

5.

Concluding remarks and future work

In this study, we proposed a methodology for describing specialized verbs and for identifying their equivalents based on FS. The methodology consisted in: extracting verbs from a comparable corpus, validating the specialized meaning(s) of the verbs, identifying their actantial structures, describing the frames that they evoke, grouping the specialized verbs together, and assigning their equivalents by means of the labels attributed to them. Every piece of information concerning the verbs served to compare the 165 pairs of candidate equivalents as well as to validate their equivalence status. The methodology revealed that the vast majority of the candidate terms indeed corresponded to valid equivalents. It also revealed that: full equivalents evoke the same frame; they have the same number of mandatory meaning slots (FEs or actants); their FEs are instantiated by means of terms that refer to the same kind of entities; and their syntactic patterns are similar. Differences concerning these elements proved to be the reason why certain specialized verbs are only partial equivalents. Taking this into account, the findings of the research indicate that the extralinguistic descriptions of the verbs (the frames that they evoke) as well as the description of their linguistic behaviour provide sufficient information to assign and examine the candidate equivalents. Moreover, the methodology also allowed us to differentiate between full and partial equivalents. Particularly helpful in the task were the actantial structures of the verbs that correspond to the core FEs of the frames that they evoke as well as the labels attributed to each frame and that were used to identify the candidate equivalents. This may be an important clue for future work, namely for the automatization of the manual process of identification described in the paper. One of the improvements to our method could consist in the automatic annotation of the contexts in which the verbs occur, such as the system proposed by Hadouche et al. (2011a) and evaluated in Hadouche et al. (2011b). As contexts contain the information used to describe the verbs and assign their equivalents, the automatization of the annotation would accelerate the terminologists work. Finally, it would be helpful to design an automatic system that groups together verbs by means of their actantial structures because, when compared to the frame labels, this kind of information is richer and more objective.

6.

Acknowledgements

We would like to thank Benot Robichaud for programming the search queries that allowed us to group together the selected verbs by means of the frame labels and that allowed us to compare the linguistic realizations of the verbs actants. We would also like to thank Marie-Claude LHomme and Rute Costa for their comments on the first version of this paper. The work described here is supported by the Fundao para a Cincia e Tecnologia, Ministrio da Educao e do Ensino Superior de Portugal (FCT-MCTES).

7.

References

Adamska-Saaciak, A. (2010). Examining Equivalence. International Journal of Lexicography, 23, pp.

1795

387--409. Atkins, S., Rundell M. (2008). The Oxford Guide to Practical Lexicography, Oxford: Oxford University Press. Baker, C. (2009). La smantique des cadres et le projet FrameNet : une approche diffrente de la notion de valence . Langages, 4, pp. 32--49. Boas, H. (2005). Semantic Frames as Interlingual Representations for Multilingual Lexical Databases. International Journal of Lexicography, 18(4), pp. 445--478. Condamines, A. (2008). Taking genre into account for analyzing conceptual relation patterns. Corpora, 8, pp.115--140. Costa, R. (2004). Corpus de spcialit : une question de types ou de genres. In H. Bjoint & F. Maniez (Eds.), De la mesure dans les termes. Hommage Philippe Thoiron, Lyon: PUL, pp. 313--224. Costa, R., Silva R. (2004). The verb in the terminological collocations. Contribution to the development of a morphological analyser Morphocomp. In Proceedings of the IV International Conference on Language Resources and Evaluation, LREC 2004. Lisbon, pp. 1531--1534. Chrom, M. (2004). Legal Translation and the Dictionary, Tbingen: Max Niemeyer Verlag. Dolbey, A.; Ellsworth M. and Scheffczyk. J. (2006). BioFrameNet: A Domain-specific FrameNet Extension with Links to Biomedical Ontologies. In O. Bodenreider (ed.) Proceedings of KR-MED, pp. 87--94. Drouin, P. (2003). Term Extraction Using Non-technical Corpora as a Point of Leverage. Terminology, 9(1), pp. 99--115. Faber, P.; Montero Martnez, S.; Castro Prieto, M.R.; Senso Ruiz, J.; Prieto Velasco, J.A.; Len Arauz, P.; Mrquez Linares C. and Vega Expsito M. (2006). Process-Oriented Terminology Management in the Domain of Coastal Engineering, Terminology, 12(2), pp. 189--213. Fillmore, C.J. (1977). Scenes-and-frames semantics, Linguistic Structures Processing. In Zampolli, A. (Ed.). Fundamental Studies in Computer Science, No. 59, North Holland Publishing, pp. 55--88. Fillmore, C.J. (1982). Frame Semantics. In The Linguistic Society of Korea (ed.). Linguistics in the Morning Calm, Seoul: Hanshin, pp. 111--137. Fillmore, C.J. (1985). Frames and the Semantics of Understanding. In Quaderni di Semantica, 6(2) pp. 222-254. Fillmore, C., Atkins S. (1992). Towards a Frame-based Lexicon: The semantics of RISK and its Neighbors. In A. Lehrer & E. Kittay (Eds.) Frames, Fields, and Contrast: New Essays in Semantics and Lexical Organization, Hillsdale: Lawrence Erlbaum Associates, pp. 75--102. Hadouche, F.; Lapalme, G. and L'Homme, M.C. (2011a). Attribution de rles smantiques aux actants des lexies verbales. In Traitement automatique des langues TALN 2011, 27 juin au 1er juillet 2011, Avignon.

Hadouche, F.; Desgroseilliers, S.; Pimentel, J.; L'Homme M.-C. and Lapalme G. (2011b). Identification des participants de lexies prdicatives : valuation en performance et en temps d'un systme automatique. In Actes de la 9e confrence internationale Terminology and Artificial Intelligence (TIA'11), Paris, France. LHomme, M.-C. (1998). Le statut du verbe en langue de spcialit et sa description lexicographique, Cahiers de lexicologie, 73(2), pp. 61--84. LHomme, M.-C. (2004) La terminologie : principes et techniques, Montral : Presses de l'Universit de Montral. Lorente, M., Bevilacqua C. (2000). Los verbos en las aplicaciones terminogrficas. In Actas del VII Simposio Iberoamericano de Terminologa RITerm 2000. Lisboa: ILTEC. Picht, H., Draskau, J. (1985). Terminology: An introduction, Guildford: University of Surrey. Pimentel, J. (2012, forthcoming). Description de verbes juridiques au moyen de la smantique des cadres, Terminologie & Ontologie : Thories et applications (Toth 2011), Annecy 2011, 26-27 mai 2011. Piotrowski, T. (1994). Problems in Bilingual Lexicography, Wrocaw: Wydawnictwo Uniwersytetu Wrocawskiego. Rogers, M. (2000). Genre and Terminology, in A. Trosborg (Ed.), Analysing Professional Genres, Amsterdam / Philadelphia: John Benjamins, pp. 3-21. Ruppenhofer, J.; Ellsworth, M.; Petruck, M.; Johnson C. and Scheffczyk J. (2010). FrameNet II: Extended Theory and Practice. http://framenet.icsi.berkeley.edu/. Accessed 12 March 2012. Sandrini, P. (1995). Der Rechtsbegriff: Implikationen fr die Terminologiearbeit. In G. Budin (Ed.), Multilingualism in Specialist Communication. Mehrsprachigkeit in der Fachkommunikation, Wien: TermNet, pp. 1105--1122. Sandrini, P. (1996). Neue Entwicklungen in der rechnergesttzten Terminologieverwaltung. In R. Arntz, F. Mayer & U. Reisen (Eds.), Terminologie fr ein vielsprachiges Europa. Proceedings zum DTT-Symposium. 85-100. Kln 11-12 April 1996. Sandrini, P. (1999). Legal Terminology. Some Aspects for a New Methodology, Hermes Journal of Linguistics, 22, pp. 101--112. Schmidt, T. (2009). The Kicktionary A Multilingual Lexical Resource of Football Language. In H.C. Boas (Ed.), Multilingual FrameNets in Computational Lexicography. Methods and Applications, Berlin / New York: Mouton de Gruyter, pp. 101--134. Songer, D. (2008). The Transformation of the Supreme Court of Canada: An Empirical Examination, Toronto: University of Toronto Press. Wiegand, H. E. (2005). quivalenz, quivalentdifferenzierung und quivalentprsentation in zweisprachigen Wrterbchern: Eine neue einheitliche Konzeption. In H. Gottlieb, J. E. Mogensen & A. Zettersten (Eds), Symposium on Lexicography XI. Tbingen: Max Niemeyer, pp. 17-57.

1796

Appendix: The frame [Regulations] groups together synonym terms, near synonym terms as well as equivalent terms Frame: Regulations Definition: Written legal documents (LAW) constitute or constrain certain matters (ISSUE). Core FEs LAW Written legal documents that have the force of law (e.g. Constitution, jurisprudence) or will have the force of law once enacted (e.g. bill, legislature). They arrange and regulate a set of Issues. ISSUE It refers to what is required or forbidden by LAW but also to rights, responsibilities and principles that are defined in the LAW. Non-core FEs:
CASE, CONDITION, MANNER, MEANS, MODE, TEXT

Contexts: Section 3(1)(a) codifies the foreign aspect of double criminality by requiring that the offence upon which extradition is requested be criminal in the requesting state and carry the specified penalty. Article 416 establishes a principle of equal partition of the patrimony between the spouses upon dissolution of the marriage. Section 84(2) governs the release of foreign nationals. The Legal Rights provisions of the Charter neither mandate nor prohibit specific conduct by Canadian officials. The Government may, by regulation, prescribe additional hiring requirements for police officers and special constables. Section 146(2)(a) provides that a statement will be inadmissible unless it is voluntary. The Act respecting administrative justice recognizes certain legal principles, such as the duty of fairness and respect for basic procedural safeguards.
O artigo 13 da Lei Fundamental consigna: 1. Todos os cidados tm a mesma dignidade social e so iguais perante a lei. O art 56 n 2 do C. P. Civil determina que no caso de execuo por dvida provida de garantia real sobre bens de terceiro a execuo seguir directamente contra esse terceiro se se quiser fazer valer a dita garantia. As normas que disciplinam a interpretao da declarao negocial so, pois, igualmente vlidas para a interpretao de uma sentena. O n2 do art.808 do Cdigo Civil estabelece que A perda do interesse na prestao apreciada objectivamente. Para alm deles, estatui o artigo 69. n.l, que: "As crianas tm direito proteco da sociedade e do Estado, com vista ao seu desenvolvimento integral... " Em sentido idntico estipula o art. 66 do C.P.Civil que so da competncia dos tribunais judiciais as causas que no sejam atribudas a outra ordem jurisdicional. Ademais, a Constituio prescreve tambm gozarem os salrios de garantias especiais, nos termos da lei (artigo 59, n 3).
LAW

Terms

ISSUE

article, chapter, legislature, provision, section codify1 Subject, NP article, court, bill, Charter of the French language, section, act, New Brunswick Human Rights Code, Criminal Code, Code, text, case law, jurisprudence, law, legislature, Ridge v. Baldwin and Nicholson Subject, NP Charter, Financial Administration Act, law, Local Authorities Election Act, Miranda, progeny, provision, regulation, rule, section, statute

aspect, case law, criterion, doctrine, immunity, principle, right, standard, test

English

Object, NP principle, rule, framework, criterion, test, scheme, duty, presumption, confine, system, right, discretion, limit, power establish1 Object, NP Complement, that-clause action, administration, admissibility, conduct, determination, disposition, joinder, management, proceedings, promotion, release, relation

govern1

1797

Subject, NP act, Charter, Constitution, Convention, law, provision, section, standard Subject, NP administrative body, City, court, Government, Governor in Council, judge, law, Parliament, section, state, statute, standard Subject, NP Taxation Act, Labour Relations Act, section, rule, Automobile Insurance Act, act, agreement, Charter, legislature, law, article, Parliament Subject, NP act, article, British Columbia Court of Appeal, Canada Transportation Act, court, doctrine, equity, jurisprudence, law, liability scheme, principle, provision, Rail Code, remedy, standard Subject, NP artigo, lei, lei preambular, notrio, propsito legal, relao Subject, NP artigo, constituio, decreto-lei, juiz de instruo criminal, legislador, lei, preceito, regra Subject, NP Cdigo do Trabalho, diploma, disposio, empregador, norma, regime, regra

mandate1

prescribe1

Object, NP approach, assessment, compliance, conduct, dismissal, floor, form, holding, inquiry, requirement, rules, use Object, NP approach, consequence, dismissal, limit, offence, period, procedure, requirement, rule, violation Object, NP case, freedom, statement, disqualification, relief, introduction, appeal, partition, compensation, rehabilitation, factor, possibility, protection, procedure, benefit Object, NP Complement, that-clause ability, burden, difference, distinction, duty, free will, judgment, obligation, presumption, principle, right

provide1

recognize1

consignar1

Object, NP agravamento, direito, modalidade, possibilidade, prazo, regra Object, NP Complement, that-clause caso, cominao, competncia, obrigatoriedade, transcrio, valor probatrio Object, NP Complement, that-clause Acidentes de Trabalho, actividade, contrato, despedimento, explorao, Grelha de Comutao, indemnizao, interpretao, extino, recurso, regime, relao, trabalho, uso Object, NP agravamento, momento, obrigatoriedade, presuno, princpio, regra Object, NP Complement, that-clause linha da forca, regime, consequencia, distino, sentido, base, principio Object, NP Complement, that-clause modo, prazo Object, NP Complement, that-clause converso, forma, princpio, tempo Object, NP Complement, that-clause

determinar1

disciplinar1 Portuguese Subject, NP artigo, cdigo, constituio, disposio, lei, n, normativo Subject, NP artigo, nmero, ECD, EDFP, Conveno Europeia dos Direitos do Homem, lei, disposio, diploma, norma Subject, NP alnea, artigo, lei, nmero Subject, NP artigo, clusula, Cdigo, Constituio, diploma, disposto, nmero, lei, redaco Subject, NP

estabelecer1

estatuir1

estipular1

prescrever1

1798