duelling theorists...

Jim and Palsberg add the syntax selftype, and deliberately choose to add this syntax, "rather than binding a name for selftype in each object type." They state that the desired meaning can always be recovered, because each occurrence of selftype comes with its context.

This bothers me, because of the following example: in Jim and Palsberg's system, what is the meaning of the type expression [makeref : [ref : selftype]] ?

I see two possible (informal) interpretations of the above type expression: either selftype represents some subtype of the entire type expression, or it represents some subtype of the type subexpression [ref : selftype] alone. I'm guessing that the former interpretation is correct, but I have not taken the time to go through the rules to verify this.

Assuming one of these interpretations is correct, is there a way to express the other interpretation? Or is this a useless goal? It seems to me that if we were to actually bind some name to each selftype for the type subexpressions, then we could unambiguously express either option.