Hi Folks,

A book [1] that I am reading says this:
-------------------------------

It is often convenient in specifying and implementing translations to treat a
translation as the composition of two simpler mappings. The first of these
relations, known as the syntactic mapping, associates with each input (program
in the source language) some structure which is the domain for the second
relation, the semantic mapping. It is not immediately apparent that there
should be any structure which will aid in the translation process, but almost
without exception, a labeled tree turns out to be a very useful structure to
place on the input. Without delving into the philosophy of why this should be
so, much of this book will be devoted to algorithms for the efficient
construction of the proper trees for input programs.

-------------------------------

The last sentence particularly caught my attention. There is a "philosophy" of
why labeled trees aid in the translation process? Fascinating! Would you
elaborate on that philosophy please? Are there articles on this?

/Roger

[1] "The Theory of Parsing, Translation, and Compiling, Volume 1 Parsing" by
Aho and Ullman, page 55.