Relative clauses and lambda calculus

Home / Computer Science


Introduction.
Basics.
Relative clauses.
Abstraction elimination.

 See also:
Plefande (a conlang)

Unlambda reference

Introduction

I've been thinking that relative clauses are just like functions in lambda calculus, and that abstraction could be eliminated using the S, K and I combinators. I thought about this when reading about the obfuscated programming language Unlambda:
http://www.eleves.ens.fr:8080/home/madore/programs/unlambda/unlambda.html

Basics

Top

Suppose we have a language where roots are isolated and the only thing one can do is "applying" a root to another root. The functor would be the modifier and the arguments modified.

BNF syntax

Syntax would be quite simple (in a kind of BNF):

construct ::= root | ` construct construct

Where ` is the apply operator. ` A B reads "apply A to B". And every other construction would be delegated on semantics.

Some simple examples

Let's work with English words as roots for some examples (I'll also assume English semantics, but it is not the point):

Intransitive verbs are just applied to nouns:

` write man
the man writes

Adjectives are simply intransitive verbs (just like in many natural languages):

` white book
the book is white = white book

Of course, application is recursive:

` write ` tall ` red man
the tall red man writes

Adverbs could be applied to adjectives or verbs:

` ` quickly write ` ` amazingly tall man
the amazingly tall man writes quickly

Transitive verbs are (in currying fashion) verbs that when applied to a nominal construction (the object, maybe?) yields an intransitive verb that, when applied to a second nominal construction (the subject), yields the expected result:

` ` write* book man
the man writes a book

Where: ` write* book corresponds to the verb "to write-a-book".

Notice that in the language, intransitive write and transitive write* should be different words (probably regularly related).

Oblique constructions should be applied to the verb, too:

` ` ` in sky fly plane
the plane flies in the sky

Many ambiguities dissapear:

` ` see ` ` with telescope man I
I see [a man with a telescope]

` ` ` ` with telescope see man I
I see [a man] with a telescope

` ` hate ` ` and dance ` eat cheese I
I hate [dancing and eating cheese]

` ` ` and ` hate dance ` eat cheese I
I hate dancing and I eat cheese
Top

Relative clauses

Top

Now let's analyse relative clauses. Just like adjectives can be interpreted like intransitive verbs or adjectives, when translated:

` red book
the book is red

` ` write ` red book man
the man writes a red book

We could do the same with verbs:

` ` eat cheese man
the man eats cheese

` write ` ` eat cheese man
the man [who eats cheese] writes

Non-trivial relative clauses

When the order is the expected, there are no problems. But what if we want to express, say:

the book [the man writes] is red

We cannot say:

` red ` ` write man book
the book [that writes a man] is red
(The book is the writer).

And neither:

` red ` ` write book man
the man [that writes a book] is red
(The sense of the relative clause is ok, but "red" should modify "book", not man).

Lambda notation

We need more expressive constructions. Adopting lambda notation for functions (where \ pretends to be the ASCII version of a greek lowercase lambda):

(\ x : ` ` write x man)

Is a construction that, when applied to an argument x, means "the man writes x".

For instance:

` (\ x : ` ` write x man) book
the man writes a book

` (\ x : ` ` write x man) novel
the man writes a novel

Therefore:

the book [the man writes] is red

Can be expressed as:

` red ` (\ x : ` ` write x man) book

Notice that if we reduce it:

` red ` ` write book man

It yields the previously discarded construction. Expressions with lambdas are not necessarily equivalent when reduced. Here, lambdas are precisely used to show that the modified root is "book" and not "man".

Top

Abstraction elimination

Top

We could invent a way of directly translating lambda expressions into our conlang.

But there is an ugly problem: the name x. In math, its said of x that it is a mute variable, as changing its name does not change the meaning of the expression.

Fortunately, this problem has already been solved with the S, K and I combinators.

Definition of the S, K and I combinators

I is the identity function:

I = (\ x : x)
For all x:
` I x = x

K is the constant function generator:

K = (\ x : (\ y : x))
For all x, y:
` ` K x y = x

And S is the substituted application function:

S = (\ x : (\ y : (\ z : ` ` x z ` y z)))
For all x, y, z:
` ` ` S x y z = ` ` x z ` y z

Lambda elimination

By induction, it's not hard to prove that any lambda expression can be reduced to an expression using these combinators, and without any mute variable.

A lambda expression (\ x : e) falls into one of these three cases:

  1. e is x

    In this case, (\ x : e) = (\ x : x) , which is precisely I!!

  2. e is a constant, or another variable not being x

    In this case, (\ x : e) = (\ x : a) , which is ` K a

    Note that: ` (\ x : a) b = a = ` ` K a b

  3. e is the application ` f g:

    (\ x : e) = (\ x : ` f g) , which is ` ` S (\ x : f) (\ x : g) (and then the we eliminate the lambda in the inner expressions f and g which, by induction, are simpler and we already know how to reduce. Note that:
    ` (\ x : ` f g) b = ` ` (\ x : f) b ` (\x : g) b = ` ` ` S (\x : f) (\x : g) b

Rules for eliminating abstraction

A rule for mechanically eliminating abstraction in lambda expressions (\ x : e):

Reading e from left to right, replace:

` with ` ` S
x with I
any constant or variable a other than x with ` K a

If more than one lambda need to be eliminated, start with the innermost ones first.

Examples

(\ x : ` red x) => ` ` S ` K red I

We should expect:
` (\ x : ` red x) book
To yield the same result as:
` red book

Let's reduce the expression:
` ` ` S ` K red I book =
= ` ` ` K red book ` I book =
= ` red ` I book = ` red book

Now let's eliminate the abstraction from:

(\ x : ` ` write x man)
=> ` ` S ` ` S ` K write I ` K man

And let's express the original problematic sentence:
the book [the man writes] is red
` red ` (\ x : ` ` write x man) book

Reducing:
` red ` ` ` S ` ` S ` K write I ` K man book =
= ` red ` ` ` ` S ` K write I book ` ` K man book =
= ` red ` ` ` ` K write book ` I book ` ` K man book =
= ` red ` ` write ` I book ` ` K man book =
= ` red ` ` write book ` ` K man book =
= ` red ` ` write book man
(Which yields, as shown before, the correct expansion but they are not equivalent in the language).

Conclusion

The power of lambda expressions guarantees that every possible relative clause is expressable in this language, and that no ambiguities can appear in nested clauses.

An example of a relative clause impossible in English from the Language Construction Kit:

This is the man [my girlfriend's father is a friend of John and him] First, let's define f (with definition as abbreviation only): f = (\ x : ` ` be ` ` of ` ` and John x friend ` ` <POSS> ` ` <POSS> i girlfriend father)
` f x = my girlfriend's father is a friend of John and x

(Notice I write the first person pronoun as "i" to avoid confusion with "I", the identity function! <POSS> is a root for possessive, where ` ` <POSS> X Y = X's Y).

The sentence using f:

` ` be ` f man this

Let's eliminate the abstraction of f:
(\ x : ` ` be ` ` of ` ` and John x friend ` ` <POSS> ` ` <POSS> i girlfriend father)
becomes:
` ` S ` ` S ` K be ` ` S ` ` S ` K of ` ` S ` ` S ` K and ` K John I ` K friend ` ` S ` ` S ` K <POSS> ` ` S ` ` S ` K <POSS> ` K i ` K girlfriend ` K father

Then, the whole sentence:

` ` be ` ` ` S ` ` S ` K be ` ` S ` ` S ` K of ` ` S ` ` S ` K and ` K John I ` K friend ` ` S ` ` S ` K <POSS> ` ` S ` ` S ` K <POSS> ` K i ` K girlfriend ` K father man this

Which is a bit quite long. Some tricks could be used to make it shorter. There's much more about it in the Unlambda reference.

These kilometric expressions make me believe this is probably violating a language universal, in the sense that it is counter-intuitive. Anyway, at least for me, it is an interesting theoretical exercise.

To create a real conlang, the apply operator should be grammaticalized somehow (as an inflection, syntactically, etc.). I leave the rest as an exercise to the conlang world.

Top
Pablo Barenbaum
Tell me what you think of this site.
Free Web Hosting