WE ARE ALL TRANSLATORS NOW : CONSTITUTIONAL ANALYSIS AS TRANSLATION

European courts and legal scholars are accustomed to construing codes that have been in place for long periods of time. In the U.S., most laws are recent enough that the meanings of their words have not changed very much over time. This, however, is not true of the Constitution, which was adopted in the late 18 century. There are debates in the U.S. about how faithful current interpreters of the Constitution should be to the original meaning of the Constitution’s language, and over what it means to be faithful to the original meaning of the Constitution’s language. Should we care about what the original drafters had in mind, or about how the public that voted on the Constitution understood the language? Scholars and judges have turned to old dictionaries for help. Now, however, corpus linguistics has entered the scene, including a new corpus of general 18 century English. In this paper, I will suggest that scholars and judges interested in the meanings of the words as then understood should put themselves in the position of lexicographers writing a bilingual dictionary that translates the terms from a foreign language Lawrence Solan: We are All Translators... 8 into contemporary English. Such a stance will bring out the many difficult problems in using a corpus as a means of making legal decisions today.

into contemporary English.Such a stance will bring out the many difficult problems in using a corpus as a means of making legal decisions today.
Legal translation has become a big industry.In 2013, the translation budget for the EU was announced to be €330 million.1Translation is a necessary part of both conducting business and resolving legal problems between legal systems that operate in different languages.When a dispute arises between a French company and a U.S. company over a contract, the relevant evidence, including the contract itself, must be available in the language of the legal system resolving the dispute.It is a necessary part of both conducting business and resolving problems that occur across a single multi-lingual legal system, such as those in Canada, Belgium, Switzerland and the European Union.In this article, I wish to explore two other circumstances that can be said to be a matter of legal translation: disputes within a monolingual legal system in which the official or working language has remained constant, but has changed over time, and more generally, the activity of legal construction as a form of translation.

Comparative Law as Translation
Comparative law has always been seen as a matter of translation, with significant discussion of terms that are similar, but not equivalent.To take a well-studied example, "consideration" in the common law of contracts relates to, but is not the same as "causa" in civilian law (for recent discussion, see Stoyanov 2016, Zweigert andKötz 1996:43).The legal world is not only multilingual but also multijural, rendering accurate translation difficult, if not impossible, when concepts in one system are not present in another.
For this reason, comparativists focus more on functional fidelity than on linguistic equivalence (Zweigert and Kötz 1996).Without taking into account the differences between the legal cultures, there really is no point in expecting concepts that exist on one legal system but not a second system to have any legitimate translation into the language of the latter.Yet, as Hendry (2014: 100) points out, "partial understanding is preferable to no understanding at all."Thus, translators do the best they can to convey at least the gist of what is being said in the source language.For example, commentators note that efforts to make English the lingua franca of the international legal community soon confront the problem that much of the terminology in English is based on common law principles, and do not carry over to civilian systems without a great deal of effort (see, e.g., Husa 2012: 174).I will not explore these issues further here because as interesting and challenging as they are, their presence is well-known.

Translation in Multilingual Legal Systems
By the same token, multilingual legal systems rely heavily various methods of seeing to it that the laws enacted in many languages all say the same thingor at least as close as possible to reaching that goal.In Canada, at least some legislation is "co-drafted," meaning that it is simultaneously drafted in both French and English, with the goal of having equivalent results regardless of which version of the law is applied in a particular case.Yet Canada's legal system itself is bijural(common law and civil law) as well as being bilingual, a fact that challenges drafters to use terminology that does not imply the predominance of one system over the other (Sullivan 2004).
Nonetheless, problems of non-equivalence arise regularly in the Canadian courts, which use the "shared meaning rule" as a default principle, generally leading to a narrow interpretation of a law whose scope is disputed.The Supreme Court of Canada describes the process: First, the English and French versions may be irreconcilable... .Second, one version may be ambiguous while the other is plain and unequivocal.The shared meaning will then be that of the version that is plain and unambiguous.Third, one version may have a broader meaning than the other.… "[W]here one of the two versions is broader than the other, the common meaning would favour the more restricted or limited meaning".2Problems quickly arise, however, when extrinsic evidence suggest that the legislature did not intend the narrower interpretation.(See Beaupré 1988;Macdonald 1997;Sullivan 2004).
The courts are aware of this possibility, and are willing to impose a teleological approach over the shared meaning rule when it appears that the narrower interpretation will undermine the intent of the legislature or the purpose of the statute.In some instances, courts may impose a "compromise" interpretation, consistent with neither the French nor the English versions.For example, a Canadian law specifies the conditions for a business paying reduced postage rates for bulk mailings.The English version of the relevant regulation referred to the "principal business" of an organization, suggesting that the rates apply only to profit-making enterprises.The French version, in contrast, applied to "l'activitéprincipale," which could be just about anything.In a 1985 case,3the court held that the rates are available only to businesses, but the businesses need not be for-profit in nature.Neither version actually carries this understanding as its literal meaning.While the shared meaning rule provides a formalism for resolving disputes over differences in meaning between language versions, the court actually adopts an essentialist approach.The goal is to uncover the core of meaning that the legislature intended to enact, and to construe the law in a manner most likely to further the law's essence.
The European Union, in contrast, relies on the legal fiction that all 24 versions of the law, each of which being both equivalent and equally authoritative to all the others, were drafted simultaneously (see Leung 2014).They were not.There is a complicated protocol spelling out which versions are to be translated into which other versions.Recognizing this hierarchy, however, would give priority to the language in which the law was originally drafted (often English, see Baaij 2015), and threaten the legitimacy of the principle that all versionsareequally authoritative.Thus, apart from early cases decided decades ago, the Court of Justice of the European Union (CJEU) almost never refers to the translation history of a law when it renders a decision that concerns differences among the language versions.Here again, translators are encouraged to use terms that do not imply the acceptance of one legal culture over others, creating a vocabulary of European legal terminology (Kjaer 2015).
Problems of legal translation in these contexts are well-studied.(See Ainsworth 2014, Kjaer 2007, 2015, Baaij 2012a and b, Leung 2014, Šarčević 2012 for recent work).Reasonably enough, the literature focuses mostly on the many difficulties that legal translators encounter in their futile quest for equivalence.To take one example from the EU,4 a directive regulated "the letting of premises and sites for parking vehicles."A Danish company, which was letting a site for boats, claimed that it was not covered by the regulation since the word "vehicles" is best understood as referring to land vehicles.Reviewing various language versions, the ECJ found no consensus.In some languages (French, English, Italian, Spanish, German and Finnish) the word seemed to apply to all modes of transport.In others (Danish, Swedish, Dutch and Greek), its most common meaning is limited to vehicles that run on land.Thus, the court resorted to the teleological approach and decided that the purpose behind the directive would be better served if boats were included within the scope of the directive (See Cao 2007: 74-75).
The case has special jurisprudential appeal because the facts closely resemble those of the famous debate between H.L.A. Hart (1958) and Lon Fuller (1958) over the circumstances in which "no vehicles in the park" applies.5Superficially, the question there is what it means for a thing to be a vehicle.However, as scholars have noted, the real issue is what should be considered a vehicle in the context of the statute, taking into account what the law was intended to cure.This inquiry requires more than reference to a dictionary.It requires inquiry into the purpose of the law, which includes an analysis of the expected uses of the parks in the community that enacts the law (Winter 2001, Eskridge 2016).
As Baaij (2012) points out there have been surprisingly few such cases confronting the CJEU over its first half century.This is not because translations are actually equivalent in all respects.Rather, the success reflects the fact that most of the time, the disputes are over issues for which the different language versions really are sufficiently equivalent.If a parking garage in Denmark had violated the directive, there would be no reason to care about the fact that the word 'vehicle' has different extensions in, say, Danish and Finnish, with respect to water vessels.Fortunately, European cultures are sufficiently similar to ensure that the prototypical examples of most legally-relevant concepts are not entirely distinct.And it is with respect to the prototypes (or "ordinary meaning" in legal parlance) that the legislation was enacted in the first place.
Thus, as a practical matter, legal translation is relatively successful, notwithstanding the many problems articulated in the literature.When language versions appear to conflict, the court examines several of them, attempting to distinguish the mainstream from the outliers, and to eliminate the latter.It further attempts to determine a core of common ground and to relate the various versions to the purpose underlying the law's enactment.I have referred to this method as "Augustinian Interpretation."Augustine recommended placing different translations of the scriptures side by side and comparing them to determine their essential message, for those who were not fluent in the languages in which they were originally written(see Solan 2009Solan , 2014 for further discussion).When the language versions diverge so there lacks a clearly superior trend, the court resorts to the teleological method, choosing the language that best appears to advance the law's purpose.Baaij (2012) notes further that the analysis is not conducted uniformly, which compromises the CJEU's adherence to an important rule of law value.Nonetheless, the comparison of different language versions often tends to reveal a common thread that characterizes most of the versions, or to demonstrate the need to look past language.
The remainder of this article looks at two seemingly monolingual situations as additional matters of translation.The first, relevant both to civilian lawyers interpreting codes and U.S. lawyers construing the Constitution, involves changes in language over time.

Constitutional Law as Translation
The U.S. Constitution is an eighteenth century document.English grammar has not changed much since then,but vocabulary has, including the meanings of words current then, and still current today.
To take an example explored by both Jack Balkin (2011)and Lawrence Solum (2015), the Constitution makes reference to "domestic violence": "The United States shall guarantee to every State in this Union a Republican Formof Government, and shall protect each of them against Invasion; and onApplication of the Legislature, or of the Executive (when the Legislature cannot be convened) against domestic Violence."6 The American Heritage Dictionary defines domestic violence as: "Physical or emotional abuse of a household member, especially one's spouse or domestic partner."That, however, is not what was meant in Article IV of the Constitution.Rather, the clause guarantees that the federal government will come to the assistance of a state that is experiencing armed insurrection against state authority.As Mark Stein (2009: 132) points out, the word "domestic"then as nowhad/has two meanings: "not foreign" and "relating to the house."If the framers wrote the clause with the first meaning in mind, and it was so understood, it would not be legitimate to switch to the second meaning in order to bring spousal abuse within the scope of this constitutional protection of state authority.
If someone unfamiliar with any of these historical facts were to hear, "the U.S. Constitution concerns itself with curbing domestic violence," it is very likely that the hearer would today (mis)understand the statement as referring to violence within the home.It is a common expression.Following earlier work of Lawrence Lessig (1993), I propose that the process by which the hearer becomes disabused of this contemporary understanding and attuned to the historical fact is a matter of translation.The claim is a strong one.The process is not like translation, it actually is translation.Just as one translates Old English into Modern English, one translates Modern English into Eighteenth Century English.
Translators make decisions.
Calling constitutional interpretation "translation" thus suggests that the kinds of decisions that legal analysts must make are the same as the kinds of decisions that translators must make.These include the legally significant concern as to whether conceptual fidelity requires that the expected applications of the term in the target language be preserved.For example, the prohibition against "cruel and unusual punishments" in the Eighth Amendment of the U.S. Constitution did not ban the death penalty in the eighteenth century.Capital punishment was clearly permitted.However, the notion of "cruelty," then as now, meant/means something like "harsher than societal norms permit."If the translator focuses on conceptual equivalence without regard to application, then changes in societal norms and fidelity to original meaning are not in tension, as theorists including Balkin (2011), Solum (2015) and Whittington (1999) point out.In contrast, the late Justice Antonin Scalia (1997: 46, 132, 145-146) argued that the availability of the death penalty in the eighteenth century ensured it a place in the arsenal of punishments permanently.
More generally, the translator must balance textual fidelity against functional fidelity.Languages do not always translate literally, requiring translator to choose words and syntactic structures in the target language that will produce the same result as the words in the source language.This is especially true when the contexts of the two differ significantly, either because of the passage of time or because of contemporaneous differences.Obvious cases include the translation of monetary language.Consider the following example discussed by Lessig (1993Lessig ( : 1176Lessig ( -1178): An 1864 law set ten dollars as the maximum fee that lawyers could charge a veteran for bringing a claim for benefits.A case arose in 1985 under that statute.Should the court have enforced the $10.00 ceiling, or adjusted it to an amount reflecting equivalent purchasing power 120 years later?The Supreme Court chose the former,7 literal path, which Lessig argues was a mistake.The purpose of the law was make sure that legal fees for such claims were inexpensive, not to bankrupt lawyers over the centuries.
Lessig's turning from a textual to a functional perspective on translation makes sense as a matter of fidelity to the intent of the drafters.The same holds true in the realm of constitutional analysis.The Seventh Amendment to the U.S. Constitution ensures the right to a jury trial in certain non-criminal cases: "In suits at common law, where the value in controversy shall exceed twenty dollars, the right of trial by jury shall be preserved, and no fact tried by a jury, shall be otherwise reexamined in any court of the United States, than according to the rules of the common law." Most discussion of this provision concern what "suits at common law" means.It is generally understood to refer to cases brought at "law" rather than "equity" in Britain at the time of the founding of the United States.But the term "dollar" is also a mystery.Since the United States was not minting money until after the Constitution was adopted, one needs to construe the word "dollar" as referring to some other currency.
American legal scholars have come to the understanding that the term "dollar" in the Seventh Amendment refers to the Spanish silver dollar, in wide circulation in the colonies at the time of the founding.One scholar (Khan 1999: 403, note 57)8describes the situation as follows: "At the time of adoption of the Constitution, Spanish, French, English, and Portuguese coins were in wide circulation and constituted the money supply of the United States.Although some gold was current, silver coins alone were legally recognized in colonial times.Of all foreign coins, the Spanish milled silver dollar was, and for decades had been, the money of American markets.The Spanish milled silver dollar was also the unit of common account.Accordingly, market transactions, exchange rates for other foreign coins, and contracts were measured in terms of Spanish dollars.As no national coins existed when the Constitution was framed, one may safely conclude that the dollar mentioned in the Constitution was a unit equivalent to the Spanish silver.The Act of 1792, which established the United States Mint, further clarified the Constitution by defining dollar as a unit of account in value and weight similar to the Spanish milled dollar".
Most legal translation, consistent with Lessig's perspective, is target-oriented.Surely, no one takes the position that lawsuits in the United States must be appraised in terms of the eighteenth century currency of Spain to determine whether the right to a jury trial exists.The goal is to make the concepts in the source document accessible and sensible to the audience reading the translation in the target language.
But this approach is controversial.Baaij (2015) has argued that source-oriented translation, even with itsawkwardness in the target language, is more transparent in that it flags for the reader passages in which nonequivalence creates conceptual difficulty.Stylistic elegance gives way to as much literal accuracy as much as is practical.Moreover, sometimes, especially in the context of very specific provisions, the word really do seem to mean what they say.Yet there are times when words simply mean what they say.The Constitution requires that an individual be at least 35 years old to serve as President of the United States.No one would suggest that because the purpose for this requirement is to ensure a certain level of maturity, that an especially mature twenty-eight year old person should be permitted to serve.
Such examples do not undermine the notion that constitutional interpretation is a matter of translation.If, say, the way we compute age had changed over the two centuries since the adoption of the Constitution (perhaps considering a newborn to be one at the time of birth, consistent with some Asian traditions), there would at the very least be legitimate debate over whether to measure age the old way or the new way.
Currently, constitutional scholars writing in the "originalist" framework argue that the words of the Constitution should be construed consistent with their "original public meaning" at which time the content of the Constitution was "fixed"(see Solum 2015).This suggests an "ordinary meaning" approach as of the time that the country was founded.The important task for the interpreter (or translator) is to determine how those who ultimately voted to ratify the Constitution most likely understood its words.Thus, it is argued, it is more important come to grips with how words were used at the time generally, than to speculate about the particular meaning that the drafters had in mind.This position, too, is controversial.Statutes are generally construed with the presumption that the legislature uses words in their ordinary sense (see Slocum 2015).This approach no doubt should apply to some instances of constitutional language as well.However, it is not at all clear that the approach should be considered the norm for constitutional analysis.Perhaps rightrs should be construed broadly as a matter of principle, and limitations on rights narrowly.As noted earlier, there is disagreement as to whether the concept, "cruel and unusual punishments" should include the list of punishments that were impermissible in the eighteenth century, or be understood as a concept whose general meaning is constant but whose boundaries may change over time (see Dworkin 1988 for discussion of the difference between concepts and conceptions in this regard).
Translation of constitutional terms across time is no easy feat.The starting point is clear enough: coming to grips with the range of available meanings in the source (older) language.However, that is just the beginning of the process.One must then decide what to do with this information.Construe the term broadly?Narrowly?Adopt an atypical usage in order to make the most sense of the language in context?If so, what contextual information should be considered?How should it be weighted?
One excellent tool used by translators and lexicographers is the linguistic corpus.The corpus, however, can only provide distributional information about word usage (see Kosem 2016).It cannot tell the legal analyst what to do with that information.Because there is no agreed upon linguistic standard for construing constitutional language, I have argued that the creation of a corpus of founding era American English, while a good idea in its own right, is not likely to solve these problems on their own (compare Phillips, Ortner andLee 2016 andSolan 2016).When it comes to constitutional law, in contrast, unless it is clear what linguistic questions we wish to ask the introduction of "big data" is not as helpful as it might be if the legal norms were uniform and wellestablished.Surely in some sense we must concern ourselves with how the language was understood at the time of the founding.However, in what sense we must do that is a contested matter (compare Balkin's2011 originalism with Solum's 2015 originalism), and the corpus cannot help us resolve this dispute.
Thus, we are back where we started.Constitutional analysis is indeed a matter of translation, but we must recognize that translators have to make decisions.Corpora are an excellent source of information about distributional facts concerning language usage.But it is not clear how much we should rely on this distributional information in constitutional analysis, and if we do, it is also not clear which distributional facts are legally relevant.While the interpretive problems are linguistic in nature, the decisions about how to resolve them are legal and political.

Extending the Paradigm
In his brilliant book, Justice as Translation, James Boyd White (1990: 236) argues that all legal analysis is a matter of translation, at the very least an extension of translation: "Interpretation is directly continuous with translation, for one who seeks to make a text in response to a text in his own language must inhabit his own version of the translator's space, knowing that it is impossible to reproduce the meaning of the prior text except in the words of the prior text, in its context, yet knowing as well that conversations can take place about such texts in which they are for some purposes, and in some ways, usefully represented in other terms".
Any time an issue in a matter of statutory interpretation leads to a statement of the form, "for purposes of this statute, the word x should be limited to situations in which y obtains," the interpreter, in essence is creating an entry n a bilingual dictionary.Of course this is more obvious when the words are in different languages, or even when they are in the same language, but separated by enough time to allow the words to have changed in meaning.Notably, as both Schauer (2015) and Durant (forthcoming) point out, many of the hardest cases of translation of this sort are instances in which legal and everyday meaning overlap, but are not identical.Translation theorists writing about multilingual legal systems make a similar point, as the example of causa and consideration discussed above illustrates.
The translation about which White speaks is creates a bridge between two linguistic cultures, reflecting different social institutions.Most frequently the bridge joins the world of legal categories and the world of everyday experience.The question, then, is not whether a bicycle is a vehicle, or even whether it is ordinarily regarded as a vehicle for purposes of deciding whether a law that bans vehicles from a public park applies to cyclists.Rather, the issue is whether, given the nature of the park and the purpose of the law, bicycle accidents are among the risks that are to be averted.Automobile traffic will be prohibited by any such law, not only because cars are prototypical vehicles, but also because they create the prototypical risk when it comes to park safety.
In many instances, however, the law defaults to the presumption that the legislature uses ordinary words in their ordinary sense (see Solan 2010, Slocum 2015, Eskridge 2016), thus making such tools as linguistic corpora of natural language especially useful in statutory interpretation (Mouritsen 2010), although at least in the United States, judges continue to rely on both dictionaries and their own intuitions about the distributional facts.The presumption that the legislature uses words in their ordinary sense not only enhances fidelity to the legislature to the extent that the presumption accurately reflects the legislative process, but it further enhances the rule of law value that the public be given fair notice of their obligations, since the presumption defaults to the understanding of the statutory language most likely to be in the minds of the public at the time of enactment.In these cases, translation is a rather straightforward task, provided that the interpreter can marshal sufficient information about how the terms are used in everyday life.

Conclusion
I have presented four different contexts in which legal actors are said to engage in translation: comparative law analysis, the functioning of multilingual legal systems, the interpretation of older documents, and ordinary interpretation of laws contemporary laws.In so doing, I do not mean to obscure the differences among these contexts.Self-evidently, there is a difference between bridging the gap between one concept in French and another in English, and bridging the gap between two contexts in which a single term is used contemporaneously.Nonetheless the similarities among these interpretive contexts are instructive, and more than metaphorical.In short, James Boyd White is correct: Legal analysis, whether across legal systems, across languages within a single legal system, across time within a single language, or across subcultures within a single linguistic environment, share a great deal.This fact certainly does not mean that the acts of translation are precisely the same in each of these contexts.It does, however, suggest that certain mental operations are essential to a variety of legal tasks, which appear superficially to be unrelated.