Jusletter IT

The notion of ‹semantic frame› has appeared in artificial intelligence related research during the 1970s. Semantic frame has been used as a concept of knowledge representation, which assumes that human understanding is based on such preliminary knowledge, which consists of representations of generalised situations or procedures (series of acts or events). Generalized situations were named frames,¹ while procedures were named scripts.² Later the use of the expression of ‹frame› and ‹knowledge frame› became common.

The hypothesis about the basic units of human understanding and knowledge representation was based on research results of cognitive psychology,³ and it has been assumed that artificial intelligence can be achieved if the necessary knowledge is represented in the form of knowledge frames (in short: frames) and if we teach machines to identify (or associate) particular real life situations or series of events with pre-defined frames (or scripts). Although the creation of such machine (or application) is far from obvious and raises a series of hard-to-answer questions, the hypothesis that human understanding happens not on the level of simple facts, but on another (maybe higher) level, which operates beyond the understanding of simple facts proved to be fruitful. The main advantage of this approach compared to others is that it defines understanding as a generic pattern recognition task, thus simplifying the problem of understanding to the recognition of pre-defined knowledge frames (patterns).

The present paper will examine the question of whether the concept of semantic frames can be utilized in the legal domain or not. Thus it revolves around the question whether we can talk about legal semantic frames or not. The paper will focus on how understanding can be understood in the legal field (what legal understanding is) and what specialities of legal understanding has compared to understanding in other fields. It will be argued the legal understanding has some specialities and as a consequence the nature of legal understanding has some important implications for the use of semantic frames in the legal domain. These implications will be shortly outlined at the end of the paper.

Today the generally accepted view about law is that it is preliminary given and comprehensible. Therefore the questions relating to understanding in the legal domain are related to legal decision making (decisions of courts or other public authorities). Legal decision making has two aspects which closely relates to understanding of the preliminary given law. One is the determination of legal facts the other is the interpretation of applicable legal rules (mostly written laws). The determination of legal facts consists of the understanding of the important (relevant) facts of the world, which consists of the comprehension of a series of acts, since past events cannot be experienced again, but only reconstructed with the help of evidences. Interpretation of applicable legal rules (legal interpretation) is evidently understanding. Legal interpretation is nothing else then discovery of meaning of applicable legal rules.

So there are two types of understanding in legal decision making. One is the understanding of evidences which underpin facts; the other is the understanding of legal rules, which are used for the legal evaluation of facts. The two types of understanding have a common element; both tasks are mainly text understanding and textual comprehension. This is because the evidences mostly exist in written form, thus the comprehension of legal facts is mainly text understanding. Legal interpretation is also text understanding, since applicable legal rules can only be found in textual form.

It is important to highlight that the determination of legal facts and legal interpretation are closely interrelated activities, thus legal decision making is often depicted as a circular (or spiral) understanding procedure.⁴ In this procedure the selection of applicable legal rules is preceded by the comprehension of the facts of the case (legal facts), which is followed by the selection of applicable rules, which is followed by the further selection of facts based on the understanding (legal interpretation) of applicable legal rules. This circular (or spiral) understanding procedure stops, when the applicability of other legal rules is not implied by the legal facts, and the applicable rules do not imply the need of observation of other facts.

This procedure obviously contains several subjective evaluative elements, because of the difference between the nature of the facts (the events of the world) and the normative nature of legal rules. Because of this difference legal understanding is not independent of the person who is in charge of legal decision making. The circular (or spiral) and subjective nature of legal understanding is properly described in the so-called synoptic legal theory of Barna Horváth. He argues that there is a mutual dependence in legal decision making of the factual and the normative and they are simultaneously present throughout the decision making procedure. In his work, Legal Sociology he summarizes this dichotomy the following way: «Selection of legal rules can happen on legal facts, the selection of legal facts happens on legal rules. A certain part of the nature (legal facts of a case) can only be understood as a legal case, if it is viewed through a legal norm (legal rule) and are selected taking into account these. Selection of legal norms and legal rules from the normative world happens with the help of their simultanous presence with legal facts: only the legal rules selected this way can be positive legal rules.»⁵

From the above we can conclude that legal understanding is an activity which relates to legal decision making, which is basically (but not exclusively) text interpretation, and its purpose is the discovery of the meaning of legal rules and evidences. Furthermore we can also conclude that legal understanding has two important aspects that comprehensive and evaluative elements are equally present in legal understanding and they are thoroughly interconnected.⁶

The understanding of legal rules and legal documents (e.g. contracts) is purely text interpretation. If we talk about legal interpretation in a narrow sense then we always refer to the interpretation of legal rules. However, there is a wider concept of legal interpretation when we include the understanding of the legal facts. Understanding of legal facts is not purely text interpretation but more and more dominantly text understanding, because more and more evidence is documentary evidence. Obviously we can also meet with scientific fact exploration methods in legal decision making procedures as well. Most legal systems allows for this purpose the use of experts (e.g. doctors, engineers) with scientific background for the discovery of the factual background of a case. These methods are obviously not text interpretation.

There is a wide literature about legal interpretation in legal theory. A general categorization of methods of legal interpretation consists of four categories: 1) linguistic, 2) logical, 3) systemic and 4) historical. This categorization stems from the work of Savigny dating back to the early XIXth century,⁷ but these distinctions can already be found in the legal practice of ancient Rome. If we look at these categories the question of why we need additional interpretive methods next to the linguistic one promptly emerges. Especially if we accept that legal understanding is mainly text interpretation and texts are linguistic expression of thoughts. The answer is surprisingly simple: it happens very often that legal cases cannot be decided on the basis of the linguistic interpretation, based on the meaning derived from the legal text using the generally accepted rules of a given language. There are two possibilities: a) linguistic interpretation does not provide guidance, b) linguistic interpretation implies consequences, which are not acceptable (e.g. because it is unrealistic). In both cases something additional is needed to reach a «proper» decision next to the linguistic interpretation. The essence of this problem is clearly shown in the following parts of the Digest of Justinian:⁸

It is also common to divide the four types of methods into two groups. The first group consist of the linguistic and the logical methods, while the second group contains the systemic and the historical. The first group is also called the lower grade of interpretation, and the group consisting of the systemic and the historical methods called the higher grade of interpretation. These expressions (lower and higher grade) indicate that linguistic interpretation must precede the others, even logical interpretation must be preceded by the linguistic interpretation, because it is impossible to draw conclusions e.g. on the contrary if do not know what is something contrary to. However logical interpretation does not need the involvement of extra textual aspects therefore this method still considered to be part of lower level methods.

The methods of legal interpretation are integral part of everyday legal practice. These are not just tools recommended by academic scholars, but their application is also required by legal rules of legal decision making. As an example Section 1 and Section 207 subsection (1) of the Hungarian Civil Code¹⁰ can be mentioned. The former states that the rules of the civil code«shall be interpreted in accordance with the economic and social order of the Republic of Hungary.” The later states that«in case of debate contractual declarations shall be interpreted in a way as the other party to the contract should have interpreted it according to the intent of the other party, the circumstances of the situation and the generally accepted meaning of words.»

Another example can be shown from the rules of the Canon Law. According to Canon 17 of the Code of Canon Law¹¹ «Ecclesiastical laws must be understood in accord with the proper meaning of the words considered in their text and context. If the meaning remains doubtful and obscure, recourse must be made to parallel places, if there are such, to the purpose and circumstances of the law, and to the mind of the legislator.»¹²

It can be asked whether the interpretation of legal texts is a specific legal method or not. In other word: is there any difference between the methods used for the interpretation of the texts of the legal documents and other texts (e.g. novels, private letters).

The answer has not just theoretical importance. If the answer is no, then we can easily conclude, that it is a wrong practice to leave the interpretation of legal texts to lawyers. We should rather employ linguists or other type of experts of humanities. Although such possibility in itself may shock lawyers, if there is no special legal element in interpreting legal texts, in this case it would make much more sense to leave legal interpretation to others than lawyers who are more skilled in such matters maybe to experts who devote their life to text interpretation or we should change the legal education accordingly.¹³

If we presume that legal understanding is legal interpretation and it is a specific method only used in legal matters then we have to find a specific legal element. Otherwise it is useless and even misleading to differentiate the interpretation of legal texts from other type of texts.

Fortunately for lawyers there is a strong empirical argument for the later to be the case. It is a common experience of both lawyers and lay persons as well that people who did not had legal education can only very hardly understand legal texts if at all. What can be the background of these phenomena?

One of the possible explanations is that legal knowledge is different than knowledge of other areas. Legal knowledge is prescriptive and not descriptive. Legal knowledge does not describe reality but it is directed toward reality. The question for legal understanding therefore is what should or shouldn’t be instead of what it is. This makes legal understanding goal oriented and provides the legal interpretation a prescriptive (normative) nature.

The goal orientedness of legal understanding implies obvious differences between legal interpretation and interpretation in other fields. The ultimate goal of legal understanding is to judge something in order to tell what should or shouldn’t be done. The understanding of a poem or a novel is not driven by the intention of this type of judgement, but by the intention of defining what should or shouldn’t be.¹⁴

This goal orientedness of legal understanding has a serious consequence for legal interpretation. The prescriptive (normative) nature of legal interpretation makes the process of legal interpretation an evaluative process. As the above cited Barna Horváth states thatlegal rules can only be selected on legal facts, and legal facts can only be selected on legal rules. This means in practice that the understanding of legal texts necessitates the understanding of the facts, and the understanding of the facts necessitates of the understanding of legal texts, and the two activities are bound together by the subject of the interpreter. This also means that the understanding of legal texts is not independent of the personality of the interpreter, his or her values, experience and his or her goal to define what should or shouldn’t be done in a given situation.

The logic, the systemic and the historical interpretation are all methods, which were developed for supporting the goal orientedness of legal understanding. They help to overcome situations where the linguistic interpretation does not provide help to define what should or shouldn’t be done. Therefore we can safely assert that legal interpretation has specialities and at least in this aspect it is an activity specific to the legal domain. Additionally in linguistic interpretation we can also find specialities like the legal definitions and legal rules of interpretation. There has been shown some of such rules in section 3.2.

From the beginning of this paper we have assumed that law is preliminary given, therefore comprehensible and thus legal understanding is nothing else then legal interpretation, interpretation of the preliminary given legal texts. This view has been the basis of the ideas outlined in the previous parts of this paper. The main message of this hypothesis is that the main function of legal decision making is the understanding of the preliminary given law. However this hypothesis is not evidently true. This hypothesis has been questioned by several people throughout the history and we can find strong arguments also in the XXth century literature, which assert that the law is not preliminary given. These arguments show that we can evidence in legal practice that legal rules are not the basis of legal decisions, thus legal rules are only used to hide the complete arbitrariness of legal decisions. These theories assert that legal decision making is not an activity which is based on the understanding and application of the preliminary given rules, but legal decision making is a creative activity. Creative in the sense that it creates the rules during the legal decision making process.

These two opposing hypothesizes (law is preliminary given – law is created in legal decision making) have been present throughout the history. Borrowing a terminology form Peter Goodrich¹⁵ we can call the first one the exegetic-hermeneutic tradition while the later one the rhetoric tradition. According to Goodrich the main characteristics of the exegetic-hermeneutic tradition are the followings: a) the tenet of doctrine,«the text is unified in advance; it represents a very peculiar and very special object of knowledge, and it is constituted as a sacred writ (something written and something sent) to handed on under very carefully controlled circumstances.» b) the tenet of legality,«The written law defines all social relations and resolves all social conflicts by reference to the legal Text.» c) the unity of meaning and univocality of language«The text has only one meaning, and the specific rule is univocal: the signifier is the neutral medium of the prior signified, the written representation of a prior speech» d) the tenet of resolution,«Law indeed is differentiated from other discourses precisely by the strictly normative character of its texts and the consequently exhaustive precognition or foreknowledge that the jurist has of legal answers to legal questions. The text provides for everything in advance; the text needs merely to be repeated and applied, comprehended and taught.»¹⁶

The rhetoric tradition however bear the following attributes according to Goddrich: a) thinking and understanding is a social activity so legal decision making is also a social activity; b) the texts and therefore its meaning«…is a series of rhetorically engaged probabilities drawn from and applicable to a specific audience» ; c) legal speech is a dialogue, and meaning is a product of a communication process.¹⁷

The two traditions are also present in modern day legal theory. Good examples of the exegetic-hermeneutic tradition are the theories of Hart and Dworkin. It is usual to present the two scholars as people who have opposing theories about law, but they very much agree that law is preliminary given. In the A Concept of Law Hart devotes a chapter to the critique of the so-called rule-scepticism, which are theories that belong to the group of rhetoric tradition according to our classification. Behind this critique we can find the concept that law is preliminary given. In other parts of his work Hart gives a clear indication of this hypothesis. He states for example:«...the theory of law as coercive orders, notwithstanding its errors, started from the perfectly correct appreciation of the fact that where there is law, there human conduct is made in some sense non-optional or obligatory.» [...people use the rules]«in one situation after another, as guides to the conduct of social life [...] For them the violation of a rule is not merely a basis for the prediction that hostile reaction will follow but a reason for hostility.»¹⁸ Similarly we can find the denial of the rhetoric tradition in Dworkin’s work Taking Rights Seriously, while he is arguing in favour of the possibility of finding the one and only right decision:«It remains the judge's duty, even in hard cases, to discover what the rights of the parties are, not to invent new rights retrospectively.»¹⁹ The two debaters seemingly agree that the law is preliminary given and comprehensible.²⁰

However the rhetoric tradition is also strongly present in the modern day legal theory. We may cite the work of Perelman and Olbrechts-Tyteca who argue in their work, New Rhetoric that rhetoric interpretation is the general method of legal interpretation.²¹ But there are less cited, but more obvious examples in the so-called realist theories (American and Scandinavian), which have been present from the beginning of the XXth century, and which have denied on various grounds that legal rules are preliminary given. The central thesis of the American relist is that law is indefinite, while the Scandinavian realist simply denied the existence of right and duties. As examples of these views we can cite Jerome Frank and Alf Ross. Frank states in his work Law and the Modern Mind:«… the widespread notion that law either is or can be made approximately stationary and certain is irrational and should be classed as an illusion or myth.» ;²² «The process of judging, so the psychologists tell us, seldom begins with a premise from which a conclusion is subsequently worked out. Judging begins rather the other way around – with a conclusion more or less vaguely formed.»²³ At Ross we can find the following in his article entitled Tû-Tû:«The claim […] obviously […] not a real thing; it is nothing at all, merely a word, an empty word devoid of all semantic reference. […] We too, then express ourselves as though something had come into being between the conditioning fact (juristic fact) and the conditioned legal consequence, namely, a claim, a right, which like and intervening vehicle or casual connecting link promotes an effect or provides the basis for a legal consequence.»²⁴

Beside the realists other schools have also been appeared after the Second World War, which attacked more directly the idea of the preliminary given law and also the role of logic (especially formal logic) in legal decision making. These schools are the so-called Critique du Droit movement, the deconstructionism represented by Derrida and legal topics which builds directly on the ancient Roman rhetoric tradition. Most important representatives of this later school are Esser, Viehweg and Perelman.²⁵

The above outlined difference between the rhetoric and the exegetic-hermeneutic tradition is very important from the point of view of legal understanding. If we presume that law is preliminary given, then there is space for legal interpretation and legal understanding. However, if law is not preliminary given and legal decision making is a creative activity, then there is no space for legal interpretation and legal understanding because there is nothing to be interpreted and understood. Not surprisingly the main message of those who follow the rhetoric tradition and especially the representatives of the deconstructionist tradition is that legal decision making is not a legal, but a political activity. If there is no preliminary given law, then we cannot see legal decision making as the operation of the pure human intellect, but only the operation of the human will, and legal decision making is only a form of exercising brute power.²⁶

In the previous sections we have identified two types of legal understanding: a) determination of legal facts and b) interpretation of applicable legal rules (legal interpretation in narrow sense). We have asserted that both types of understanding are dominantly text understanding. We have also argued that legal understanding is specifically legal activity, because legal understanding is goal oriented activity. We have also looked at the possibility that legal understanding is not a comprehensive but creative activity. From these we can draw three conclusions regarding the possibilities of using semantic frames for the purpose of machine understanding in the legal domain.

First of all we can conclude that there is a wide range of possibilities to use semantic frames in relation to legal decision making, because legal decision making necessitates the understanding of evidences and legal rules.

We can also draw a conclusion in relation to the possibility of the creative nature of legal decision making. We have to accept that the presence of the human will in legal decision making can render the understanding of legal rules unnecessary. If no understanding is needed then there is no use of any tools supporting understanding. Creating or using semantic frames in the legal domain have to take into account this.

A third conclusion relates to the circular (spiral) nature of legal understanding. This makes it impossible to simplify legal understanding process to a single act. As we have seen facts are selected on rules and rules are selected on facts during legal decision making. Therefore legal understanding is not a process which is directed toward a pre-defined goal. It is more similar to a conversation among the decision maker, the legal rules and the evidences. Therefore applications which would like to support legal decision making may be more useful if they are able to support this dialogue instead of offering an ultimate solution to understanding. This is an advice to any type of legal decision support tools not a specific requirement which only applicable to semantic frames.

Baker, Fillmore and Lowe , The Berkeley FrameNet project, in. Proceedings of the 17th international conference on Computational linguistics – Volume 1, (1998).
Betti , Emilio, Az értelmezés általános elméletéről, in. Jog és nyelv, Budapest (2000).
Bódig, Mátyás , Hart, Dworkin és a jogelmélet posztmetafizikai fordulata, Osiris, Budapest (2000).
Brósz and Pólay , Római jog, Nemzeti Tankönyvkiadó, Budapest (1974).
Dworkin , Taking Rights Seriously, Harvard University Press (1997).
Erdő , Egyházjog, Szent István Társulat, Budapest (2005).
Fillmore , Frame semantics and the nature of language, in. Annals of the New York Academy of Sciences, New York (1976).
Fillmore and Baker , Frame Semantics for Text Understanding, in. In Proceedings of WordNet and Other Lexical Resources Workshop, NAACL, Pittsburgh, (2001).
Frank, Jerome, Law and the Modern Mind, Transaction Publishers, New Brunswick N.J. (2009).
Futó (ed.) , Mesterséges Intelligencia, Aula, Budapest (1999).
Gawron , Frame Semantics, http://www.hf.uib.no/forskerskole/new_frames_intro.pdf, last, accessed 30.4.2010 (2008).
Goodrich, Peter, Historical Aspects of Legal Interpretation, Indiana Law Journal, Vol. 61, No. 3, pp. 331-354 (1985-1986).
Hart, H.L.A. , The Concept of Law, Clarendon Press, Oxford (1997).
Horváth, Barna, Jogszociológia, Osiris, Budapest (1995).
Karácsony, András, Gadamer és Betti vitája a jogi értelmezéstanról, in. Jogelméleti Szemle, Vol. 2000, No. 2, http://jesz.ajk.elte.hu/karacs2.html, last accessed 30.4.2010 (2000).
Karácsony, András, Jogfilozófia és társadalomelmélet, Pallas Stúdió, Budapest (2000).
Minsky, Marvin, A Framework for Representing Knowledge. The Psychology of Computer Vision, in. P.H. Winston (ed.), McGraw-Hill, (1975).
Petruck , Frame Semantics, http://citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.4.7422, last accessed 30.4.2010 (1996).
Ross, Alf, Tû-Tû, Harvard Law Review, Vol. 70, No. 5, pp. 812-825 (1956-1957).
Samu and Szilágyi, Jogbölcselet, Rejtjel, Budapest (1998).
Schank and Abelson, Scripts, Plans, Goals and Understanding, Hillsdale, (1977).
Searle, John, Elme, nyelv és társdalom, Vince, (2000).
Szabó, Miklós, Szó szerint... A jog és nyelv interferenciájáról, in. Jog és nyelv, Budapest (2000).
Szilágyi, Péter, Jogi Alaptan, Osiris, Budapest (2001).
Venturi, Lenci, Montemagni, Vecchi, Sgri, Tiscornia and Agnologni, Towards a FrameNet Resource for the Legal Domain, (2009).
Vámos , Knowledge and Computing – A Course on Computer Epistemology, CEU Press, Budapest – New York (2010).
Varga, Csaba (ed.) , Összehasonlító jogi kultúrák, Budapest (2000).
Villey , A joglogika története, in. Csaba Varga: A jogi gondolkodás paradigmái, Budapest, (1998).
Visegrády, Antal , A jog és állambölcselet alapjai, Dialog Campus, (2001).
Watson, Alan, The Digest of Justinian, University of Pennsyslvania Press, Philadephia (1998).

---

Balázs Rátai, Memeber of the ICT Research Centre, Universiyt of Pécs, Faculty of Law, 48-a tér 1., 7622 Pécs, HU,balazs.ratai@carneades.hu ;www.carneades.hu ,www.ikjk.hu

---