The Interpretation/Construction Distinction Revisited

You probably recognize the “interpretation/construction distinction” as a crucial part of the “New Originalism,” at least as presented by the likes of Keith Whittington, Randy Barnett, and Larry Solum.  The concept, as I understand it (and my understanding, or lack thereof, may be the real problem here) is something likes this.  When interpreting legal texts, we go through a two-step process: (1) We interpret the language’s bare “semantic meaning”; (2) We then use this semantic meaning as a cornerstone in constructing the language’s “legal meaning.”  In the first step, we are concerned with fairly straightforward problems like textual ambiguity and anachronism.  In the second step, we must resolve any residual semantic vagueness by constructing a determinate legal meaning. 

A year or so ago I contributed to a discussion on this topic in an issue of Constitutional Commentary, and I tried (rather ham-handedly, in retrospect) to explain that, while I think that the distinction exists and fairly describes a limited class of interpretive processes, ultimately it can do very little theoretical work in the controversial cases that are of most consequence.   In that piece I mentioned Willard Quine’s attack on the dogmas of empiricism in passing, but I failed to make it a central part of my critique.  In doing some new work on Quine, however, I have come to wonder whether my real issue with the interpretation/construction distinction is that it falls prey—at least in roughly analogous ways—to the Quineian critique of the analytic-synthetic distinction.  I’ll try to explain below, hopefully with less hammy hands.

First, let me give an abbreviated take on Quine and the analytic-synthetic distinction.  Analytic statements (going back at least to Kant) are thought to be those that we can assess as either true or false based simply on the meaning of their terms.  The classic example is “All bachelors are unmarried men.”  Synthetic statements are those that require us to do some empirical and/or theoretical work to verify or falsify.  In some sense, analytic statements are the a priori truths we can use to synthetize more elaborate views—they are the “foundation” in some sense of our later buildings.  Quine pointed out, however, that analyticity (at least in informative contexts) depends on the concept of synonymy, which itself requires at least some theoretical constructs and some empirical work.   We have to disambiguate different kinds of bachelors for one (degrees vs. men-about-town), and we have to have some at least basic theory of linguistic reference already in place.  Quine concedes that analytic statements exist—and that they are conceptually distinct from synthetic statements—but contends that the distinction gets very porous when we move beyond the most basic kinds of statement.  We can all agree, for example, that “All bachelors are bachelors” is analytic—but this doesn’t get us very far.  In other words, to say that anything beyond “A=A” is analytic is to embed and understand a statement within some preexisting theoretical context—and this ultimately makes the distinction quite fuzzy.

 Now, let me move on to the interpretation/construction distinction.  I think (though I could be wrong) that most New Originalists concede that “semantic meaning” (which we interpret) is not something analytic in the sense I’ve described above.   They happily acknowledge that we should do things like disambiguate terms and avoid anachronism (“domestic violence” doesn’t mean what it once did) in the interpretation phase, which means that we must have some theoretical pieces already in place as we “interpret”.    As a very simple example, there would be no need to clear up anachronism if we didn’t already think that “semantic meaning” should be fixed at the time of utterance, or something like that.  Further, for purposes of the distinction, its proponents are generally agnostic about the theoretical choices that should guide either the interpretation or construction phases.   But…

In the context of a larger originalist approach—as a piece of the apparatus, so to speak—I think the “interpretation” phase is meant to reveal something (“semantic meaning”) which functions in roughly the same way that analytic statements do in logical positivism.  After some basic moves, made in light of broadly accepted theoretical commitments, we arrive at a semantic meaning based on something like “linguistic facts.”   This meaning is, in theory, something that for practical purposes is objective or verifiable on the face of the language itself (given some set of “linguistic facts” that, again, we can theoretically verify).    With this bit of semantic foundation in place, we can go about the more difficult business of “constructing” (or synthesizing) determinate legal meanings when the semantic meaning is problematically vague.  As Randy Barnett is fond of saying, “There’s a there there”—and what’s “there” is the language’s semantic meaning.  So far, so good.  The interpretation/construction distinction seems to be doing much of the same work the analytic/synthetic distinction was once thought to do.

 The problem, however, is that for interpretation to work in this way—as revealing something basically a priori—one of the theoretical constructs that must be in place is an account of linguistic reference and meaning.  And the account that the New Originalists rely on (roughly Gricean?) simply is not accepted broadly enough.  In fact, I think it’s wrong.  This might not be too much of a problem, except that the account of meaning that Iaccept is something much closer to that which the later Wittgenstein gave.   Roughly, according to that account words often don’t actually refer to anything, either in the world or in our minds.  They derive their meaning from their proper use in a given context—in other words they point us towards linguistic rules, not facts.  Solum, for one, seems to argue that we can discover these rules as if they were facts, and then simply apply them in the interpretation phase.  I think that this fundamentally misunderstands what we mean by linguistic rules. 

 Linguistic rules are in fact imprecise and changing.  There is no necessary or sufficient set of conditions that define when a word is used correctly (ie, its meaning).  Rather the rules map out what Wittgenstein famously called the “family resemblance” a word bears to itself across various contexts (he asked us to consider all the acceptable uses of “game” as an example).   Thus, the rules themselves require quite a bit of interpretation and construction, all done in light of fairly specific contextual conditions. Thus they cannot, I don’t think, serve as “facts” in the ways that Solum hopes.

 More to the point, if we understand meaning in these terms, it starts to look like an awful lot of theory has to be in place and agreed upon before we can arrive at “semantic meaning” in all but very basic kinds of cases—those where the linguistic rules are so clear that they might serve actually as something like facts.  With all of this theorizing and empiricism going on in the interpretation phase, it starts to look at lot like the construction phase, where we are working to iron out textual vagueness.   The borderline between the two phases has become every bit as porous as that Quine posited between analytic and synthetic statements.  

 To put it perhaps more simply, words are always vague, some just more than others.  And before we can construct determinate “legal” meanings in the face of vague “semantic” meanings, we must first construct determinate “semantic” meanings in the face of vague linguistic rules.

 Phew.  That turned out to be a tome, and perhaps more hammy than ever.  I apologize…but if you got this far, I’d love to read your thoughts.