Comment author: @lpw25

See section 4.1.1 of the manual.

Comment author: @gasche

More precisely, the "Function Applciation" documentation
http://caml.inria.fr/pub/docs/manual-ocaml/expr.html#sec120
explicitly mentions that the application form is "f arg1 arg2 arg3 ...", so the semantics of "(f x y) z w" and "f x y z w" need not be the same (the former has two forms, the latter only one, it's a different AST). The textual explanation details the optional parameter semantics in a way that explains the difference you observe.

See Jacques Garrigue' research work with Jun Furuse on optional parameters for a formalization of the labels as found in OCaml:

http://www.math.nagoya-u.ac.jp/~garrigue/research.html#sel

A label-selective lambda-calculus with optional arguments and its compilation method
Jun Furuse and Jacques Garrigue, 1995
http://www.math.nagoya-u.ac.jp/~garrigue/papers/rims-1041.html

Comment author: bay2@illinois.edu

Gasche,

Thank you for your answer. I understand the application form as f arglist.

However, the example seems to give lie to the statement from the manual:

The expressions ( expr ) and begin expr end have the same value as expr.

Comment author: @gasche

I see what you mean: he fact that "(expr)" and "expr" evaluate in the same way is not compositional: it does not mean that "(expr) foo" and "expr foo" evaluate in the same way. But this is true in most languages, for example "(1+2)3" and "1+23" are distinct, yet explaining that parentheses do not affect the value seems reasonable even of arithmetical expressions.