RE: Of Legal Language and CS Notation

["Anton van Straaten" <address@hidden>]

Peter J. Wasilko wrote:
> > (c) Many contracts could be expressed more clearly if they could use
> > the constructs of very basic computer programming to a greater extent
> > than they can already; the main reason they can't is (a).
> >
>
>     This is very true. We actually tried to use programming to improve
> our reasoning in one experiment when I was in law school. The idea was
> that if we taught a bunch of law students to model their legal briefs as
> Prolog programs all of their logical lapses would be manifest by the
> system.
>
>     Needless to say, the law students had a hard time going from English
> to Horn clauses, didn't grok unification, and pretty much lost it when
> cut was introduced.

Along these lines, there seems to be something of a movement to use lambda
calculus in analyzing natural languages.  For example, here's an upcoming
workshop on the subject:
http://www.dcs.kcl.ac.uk/staff/maribel/Workshop-Kings.html
(Note the type theory connection, also.)

Here's an intro to lambda calculus with a linguistic bent:
http://ling.ucsd.edu/~barker/Lambda/

Googling for terms like "lambda linguistics" turns up plenty more.

There are some obvious and easy applications of lambda calculus to something
like a legal contract: for example, variable binding and lexical scoping
would allow things like a clear definition of the scope of terms, and the
ability to parameterize text precisely.  Not as ambitious as detecting
logical errors, but perhaps still better than the state of the art.  I've
written code to manipulate legal contracts, so I have some idea of how loose
they can be in terms of structure, compared to a more formal expression.

In short, perhaps law schools should start out experimenting with something
like Scheme, rather than Prolog...

Anton