This paper deals with the problem of reasoning with synonymic expressions in the domain of statutory law. It is shown that, even in cases of strong lexical synonymy (what is referred to here as 'Almost Identical Expressions'), it is necessary to engage in complicated argumentative structures in order to obtain justified conclusions concerning the mutual interreplaceability of legal terms. This result has implications for the methods adopted in research on the automated analysis of the corpora of legal texts.
The aim of this paper is to analyse the phenomenon of legal argumentation that makes use of almost identical expressions extracted from statutory texts. It is often the case that a lawmaker makes use of expressions A and B in statutory rules, where A and B have such a similar meaning that they would be presumably treated as strictly synonymic by a native speaker of a language. Therefore, a native speaker of a language would be inclined to assign identical consequences to states of affairs designated by expressions A and B. The similarity between the mentioned expressions seems to constitute grounds for the application of arguments based on plain meaning and analogy. However, there exist certain rationality assumptions behind the making of laws leading to the conclusion that, if two expressions are not strictly identical, they should be treated as different by the addressee of the statutory regulation. These two argumentative stances point out inconsistent solutions and, therefore, may cause divergent opinions concerning the rights and obligations of addressees of the law. Therefore, the investigation of this phenomenon is important for the sake of legal policy matters. However, the analysis of argumentation encompassing almost identical expressions is also of crucial importance for the development of legal knowledge-based systems. Such systems should take into account that the relation of synonymy between linguistic expressions should be treated more carefully than in less formal contexts of discourse in order to avoid oversimplifications and potentially wrong suggestions to the user. 2
Synonymy has always been considered one of the
most basic semantic relations between linguistic
expressions
Due to these problems, the relation of
synonymy has been a subject of interest for linguistic
philosophers. A classic contribution to the debate
is a paper by Goodman (1949), in which he argues
than no two non-identical words can have the
same meaning. Instead of the theory of synonymy
1 The WordNet project involves the notion of synsets: sets
of cognitive synonyms that represent certain concepts.
http://wordnet.princeton.edu, accessed on September 24,
2014.
as ‘sameness of meaning’, he advocated a theory
of ‘likeness of meaning’, according to which two
names of predicates may be treated as synonyms
if their meaning is similar enough to warrant the
thesis of their ‘sameness’, or mutual
interreplaceability, in certain contexts of discourse. The
criteria used here may vary from one context to
another
As with any philosophical thesis, Goodman’s
proposal remained controversial in the literature
of the subject
Philosophical controversies notwithstanding, the notion of synonymy is widely used in lexicography, and the existence of thesauri and dictionaries of synonyms is obvious evidence for the usefulness of this relation for language users. The words ‘synonym’ and ‘synonymy’ are actually used by the speakers of languages, and the corpora of conversational material are investigated in order to establish their actual understanding of the term. Murphy (2013) notes the following accounts of the word ‘synonym’ as found in the analysed corpora:
1) synonymy as ‘sameness’ or ‘near sameness’ of meaning,
2) synonymy as the possibility of substituting one word for another,
3) synonymy as the co-extensional character of two scientific names (in biology).
There are more specific understandings of the
word ‘synonym’ in computer science
It is easy to note that the use of the term
‘synonymy’ in descriptive lexicography tends to avoid
the discussion of philosophical problems of this
linguistic phenomenon. Generally, the people
interested in finding synonyms to certain words are
interested in substitutability of these words
without changing their meaning
The texts of statutes consist of linguistic
expressions. Generally speaking, a lawmaker intends to
indicate certain states of affairs and to assign legal
consequences to them. The lawmaker indicates
these states of affairs by means of linguistic
expressions. The language of law shares many
features with plain language, such as indeterminacy
and vagueness (Bix, Endicott); however, although
it is often presumed that statutory texts should be
understood with regard to the ‘plain, natural
meaning’ (Interpreting Statutes), often special,
legal meaning should be ascribed to the used terms
It is often the case that the lawmaker chooses similar, yet not identical, terms to refer to certain states of affairs that are assigned to legal consequences. In such contexts, there is a situation of doubt whether the lawmaker intended to refer to the same, or to different (sets of) states of affairs. The pragmatic context of interpreting such statutory language expressions is set out by the adversarial character of legal proceedings. Each party is interested in persuading the judge to ascribe such meaning to a statutory term that leads to the legal consequences desired by this party. Consequently, a party to the dispute may be interested in treating similar expressions alike with respect to their legal result; another party may be interested in strict differentiation between the meanings of slightly different expressions.
There are different approaches to the indicated problem in different jurisdictions. Sometimes, even the lawmaker gives explicit guidelines to show how similar expressions should be interpreted. For instance, the Australian Acts Interpretation Act 19012 contains a provision, 15AC, according to which, ‘when an Act has expressed an idea in a particular form of words and a later Act appears to have expressed the same idea in a different form of words for the purpose of using a clearer style, the ideas shall not be taken to be different merely because different forms of words were used.’ 3 However, typically, the lawmaker will be reluctant to give the addressees of legal texts such explicit suggestions. Thus, the dilemma concerning the ascription of identical or non-identical meaning to slightly different linguistic expressions will remain an open issue.
This dilemma is particularly visible with regard to the class of expressions we refer to as Almost Identical Expressions (AIE). By definition, the linguistic expressions E1 and E2 in language L belong to the set of AIE if and only if: 1) they stem from the same lexical root, 2) they are not identical from the syntactic point of view,
3) they would be considered as natural mutual substitutions by a competent native speaker of language L (in a relevant context of discourse C).
The point 3) is the most important one: AIE create a strong inclination in the native speakers of the language to treat them interchangeably in the relevant context of discourse. But point 2) creates the possibility for the construction of arguments to the contrary. The next two sections are devoted to the discussion of an exemplary legal question encompassing the use of AIE. 4
The legal research problem that focused our attention on the argumentation concerning AIE is as follows: what are the legal consequences of noncompliance of subjects of law with the requirement of concluding contracts and making other statements in writing? The Polish Civil Code (Act of 23 April 1964, consolidated version: Journal of Laws 2014.121, hereafter referred to as the PCC) contains approximately 100 instances of expressions lexically cognate with the word ‘writing’, most of which are parts of provisions specifying requirements of the form of contracts and other statements. There are three types of these expressions, forming a set of AIE: 1) ‘in written form’ (PL: w formie pisemnej), 2) ‘in writing’ (PL: na piśmie) and 3) ‘stated in writing’ (PL: stwierdzone pismem). 3 We are grateful to Graeme Hirst for pointing out this interesting regulation during the BiCi seminar on Frontiers and Connections between Argumentation Theory and Natural Language Processing in Bertinoro (July 20-24th, 2014).
All these expressions would be treated as mutual substitutions in the majority of contexts of discourse by a native speaker of the Polish language; interestingly, lawyers are also often inclined to see these three expressions as interreplaceable ones. However, this contention does not lead to any immediate answers concerning both the content of requirements that are expressed by the expressions listed above and the consequences of non-compliance with these requirements.
For the sake of clarity regarding the following investigations, it is necessary to delineate the legal context concerning the ‘written form’ requirement in Polish civil law. The basic rules dealing with this issue are in art. 73 § 1 of the PCC:
If the law stipulates that a legal act be made in written form, an act made without observing the stipulated form is invalid only if the law provides for a nullity clause.
and article 74 § 1 of the PCC:
The stipulation of written form without a nullity clause leads, if the stipulated form is not observed in litigation, to witness evidence or evidence in the form of declarations of the parties concerning the performance of the act being inadmissible.4
The legal consequences stemming from the quoted rules are straightforward. If a given act should be made in written form and the law prescribes for the pain of nullity, in the case of failure to fulfil the requirement, the act is not valid. Conversely, if the pain of nullity is not mentioned in the law (or in the statement of the parties), the act cannot be invalid in the case of non-compliance with the written form requirement. This consequence is uncontroversial. The legal results provided by the latter of the quoted provisions are more nuanced: if a written form is required for an act and it is not complied with, the act is still valid. However, certain types of evidence are not admissible to prove that such act has taken place. Let us refer to this legal consequence as the consequence of evidentiary difficulties. In the following analyses, we will focus on this latter legal consequence only. The consequence of invalidity is an easy topic from the point of view of argument 4 The translations of the provisions are taken from the commercial Legalis system provided by the C.H. Beck publishing house, with certain modifications by the authors. mining and natural language processing of statutory texts: an act is invalid only if there is an explicit clause providing for such consequence. In the absence of such a clause, the consequence of the failure to meet the requirement of a ‘written form’ should lead to evidentiary difficulties. This contention is, again, uncontroversial, with regard to the requirement of ‘written form’ as indicated in the latter of the quoted provisions. The question is, first, whether the requirements provided by the law should be understood identically where the law speaks about ‘written form’, ‘in writing’ and ‘stated in writing’, respectively. Second, what are the legal consequences of the failure to meet the requirements referred to as ‘in writing’ and ‘stated in writing’?
Let us present the existing controversy in a more explicit manner. Let us assume that a legal provision of the PCC has the following scheme: (X) Legal act X should be performed in written form.
The quoted art. 74 § 1 of the PCC enables us to derive the following conclusion from (X): (X-con) If the legal act X is not performed in written form, then the consequence of evidentiary difficulties shall apply as regards the legal act X.
Let us recall the expression ‘in written form’ forms an AIE set with the expressions ‘in writing’ and ‘stated in writing’. This enables us to present the two following schemes of provisions (actually often instantiated in the PCC): (Y) Legal act Y should be performed in writing. (Z) Legal act Z should be stated in writing.
The precise formulation of the legal research questions goes as follows: (Q1) Is the meaning of X, Y and Z identical? (Q2) Is it the case that Y and Z lead to the formulation of Y-con and Z-con rules analogous to the X-con rule?
In order to establish valuable answers to these questions, a corpus of judgments (>30 cases) and legal doctrinal works (5 sources) were examined. The results are reported in the following section. 5
The analysis of the existing material led to the following answers to the questions outlined above: Q1: undecided (there are authoritative sources that tend to give positive and negative answers to the question) and Q2: positive (but the interpretation of the answer depends on the chosen answer to Q1).
Theoretically, several argumentation schemes can play their role is justifying different answers to Q1. For instance, the argument from plain natural meaning would support a positive answer to Q1. The argument would run as follows.
Premise 1. Statutory terms should be interpreted in accordance with their plain natural meaning.
Premise 2. According to plain natural meaning, the expressions ‘in written form’, ‘in writing’ and ‘stated in writing’ should be treated as (strict) synonyms.
Conclusion. The meaning of X, Y and Z is identical (positive answer to Q1).
Let us note that this argument could be further backed by analogous reasoning: Premise 2 could be refined to relax the assumption of strict synonymy in favour of the claim that, in the context if legal discourse, these AIE should be treated as carrying the same meaning (because the differences between them could be reasonably ignored).
Actually, a refined version of this argument
scheme was used by one of the most influential
legal scholars in Poland,
Premise 1. If differences between the terms used by the legislator are a matter of style only, then the terms should be treated as (strict) synonyms.
Premise 2. ‘In written form’, ‘in writing’ and ‘stated in writing’ are terms that differ with respect to style only.
Conclusion. The meanings of X, Y and Z are identical (positive answer to Q1).
Let us note that a positive answer to Q1 implies, as a matter of logic, a positive answer to Q2.
However, it is also possible to formulate arguments to the contrary. According to the rationality postulates concerning legislative process, if the legislator intends to indicate the same state of affairs in different parts of regulation, he uses one and the same term. If he uses (even slightly) different terms instead, this means that his intent was to designate different states of affairs. This argumentative pattern is often referred to as the prohibition of synonymic interpretation:
Premise 1. The terms used in the statute should not be assigned with an identical meaning unless they are syntactically identical.
Premise 2. ‘In written form’, ‘in writing’ and ‘stated in writing’ are not syntactically identical.
Conclusion. The meanings of X, Y and Z are not identical (negative answer to Q1).
Note that a negative answer to Q1 does not logically imply a negative answer to Q2. A negative answer to Q1 consists only of holding that the ‘written form’ requirement is something other than ‘in writing’ or ‘stated in writing’. Let us add in this connection that, uncontroversially, the ‘written form’ requirement is satisfied only if a statement is manually5 undersigned by a person.
Consequently, the controversy between a positive and negative answer to Q1 boils down to the set of sufficient conditions to satisfy a given requirement. Undoubtedly, if a legal provision is based on the scheme (X) presented above, the requirement is not met unless the statement encompassing the content of legal act X is manually undersigned by a person. The question (Q3) is whether this sufficient condition should also be met for the satisfaction of requirements formulated in schemes (Y) and (Z). As a matter of course, a positive answer to Q1 implies a positive answer to Q3, while a negative answer to Q1 implies a negative answer to Q3.
Interestingly, the judicial opinions reviewed in the research tend to adopt a rather negative answer to Q1 (unlike doctrinal sources quoted above). This may be caused by the fact that judicial authorities are closer to legal practice and they do not intend to impose unnecessary burdens on the addressees of the provisions. This is particularly visible in the context of the interpretation of the following provision (art. 514 of the PCC) related to the institution of a claim assignment:
If a claim is stated in writing, a contractual stipulation that assignment cannot be made without the debtor's consent is effective towards the assignee only when the document contains a mention of the stipulation unless the assignee knew of the stipulation at the time of assignment. 5 For the sake of brevity, we leave the problems of electronic signatures aside.
The courts tend to adopt a negative answer to Q1 in this context. For instance, in the Resolution of 6 July 2005 (III CZP 40/05), the Supreme Court stated that:
Stating of the claim in writing in the understanding of the art. 514 of the PCC is satisfied also in case the creditor issues a document (e.g. an invoice) that confirms the performance of an obligation and the debtor accepts the document.
Thus, the Supreme Court ruled that the requirements for satisfying the ‘stated in writing’ requirement are less severe than ‘in written form’. The satisfaction of the latter implies the satisfaction of the former, but not the other way around.
The reconstruction of an argument justifying this conclusion from the wording of the Resolution is a non-trivial task due to the highly complex structure of the analysed sentences. The proposal of the argument’s structuration would be as follows:
Premise 1. There is no need to delimit the types of documents that may be used for the identification and confirmation of legal facts (wrt art. 514 of the PCC).
Premise 2. Adoption of a positive answer to Q1 would amount to the undue delimitation of the types of documents used for the identification and confirmation of legal facts.
Conclusion. Q1 should be answered negatively.
The argument formulated by the Supreme Court is enthymematic, especially with regard to the premise 1: the court seems to assume that the possibility of identification and confirmation of legal facts is a worthwhile value, which should be realised at the expense of more firm protection of debtors. This stems from the contention of the Supreme Court, according to which an invoice issued by the creditor but not accepted by the debtor would be insufficient to fulfil the condition of ‘being stated in writing’, because the protection of the debtor would be too weak if a broader interpretation were accepted. This value judgment can be reconstructed from the text only by a person who possesses at least basic legal training. However, this does alter the conclusion that the Supreme Court rejects the thesis concerning the mutual interreplaceability of expressions ‘in written form’ and ‘stated in writing’.
It is worth emphasising that the same interpretation has been accepted by the courts with regard to the interpretation of art. 511 of the PCC:
If a claim is stated in writing, its assignment should also be stated in writing.
For instance, in the Judgment of the Appellate Court in Katowice of 8 March 2005, I ACa 1516/04, the negative answer to the Q1 was advanced on the basis of a literal reading of the statute: If the legislator speaks about ‘stating in writing’, this means that he does not intend to introduce a requirement of ‘written form’, simply because these expressions are not identical.
Let us note that the answer to the Q2 may remain positive even if Q1 is answered negatively. However, different situations will have to be considered as regards the satisfaction of ‘written form’ and ‘stated in writing’ requirements. 6
The investigations of this paper lead to the formulation of the following conclusions. The peculiarities of statutory text make the NLP analyses related to this material very difficult. In particular, such ubiquitous semantic relations as synonymy have to be dealt with in a non-standard manner as regards the statutory text. Even in the case of AIE that seem to be very close, or even perfect synonyms in other contexts of discourse, establishing the interreplaceability relations between terms is a problematic issue. Reaching a justified conclusion as regards this relation in legal contexts is a complicated process, also due to the fact that lawyers disagree about the existence or non-existence of synonymy relations between the analysed terms. This process involves the reconstruction of legal arguments used in different authoritative sources. The reconstruction is not an easy task due to the complicated structure of sentences present in judicial opinions and doctrinal theories as well as posing hypotheses about enthymematic premises. The latter activity involves a vast amount of professional legal knowledge. Therefore, the corpora of legal texts should be annotated by legal professionals (or at least legal students) in the process of argumentation mining rather than by laymen in order to avoid misunderstandings generated by a lack of legal knowledge.
Even in the case of AIE, which seem to be (near) synonyms on purely linguistic grounds, as it was shown, the discussion of their interreplaceability involves the use of not only linguistic arguments, but also teleological arguments possessing a complicated structure. The obtained conclusions are contextual and perhaps defeasible, as is often the case in the context of legal discourse.
The most important conclusion stemming from the investigations above is that, in the context of an NLP analysis of the corpora of legal texts (aiming at the creation of intelligent databases of legal knowledge), one should be very cautious as regards the use of any databases of synonyms. Moreover, the corpora of statutory texts should not be analysed apart from the legal doctrine and (most importantly) databases of legal cases. These sources should serve for the reconstruction of arguments used to determine the meaning and scope of statutory expressions.