Please see my comment to the question, other comments. It's not feasible to answer this question in any practically sensible way.
However, I can formally work around the situation. From that standpoint, the answer is: the
Java compiler does not really "interpret" those words. After scanning and parsing (optimization, etc), it merely compiles the code to Java
bytecode, pretty formally. A compiler does not "know" the semantic of the code. Compiled Java source code is not a machine code, it is a cross-platform bytecode.
Next step is done by a JIT compiler:
http://en.wikipedia.org/wiki/Java_bytecode[
^],
http://en.wikipedia.org/wiki/Just-in-time_compilation[
^].
See also:
http://en.wikipedia.org/wiki/Bytecode[
^].
-SA