Why do lexical rules for keywords conflict with identifiers even if I specify the special case (keywords) before the more general identifier rule?

What is wrong with the following? Doesn't the lexer know by the order in which I specify the two tokens that I want BEGIN to be a special case of an ID?

class TestLexer extends Lexer;
...
BEGIN	:	"begin"
	;

ID	:	('a'..'z')+
	;

There is no order dependency between the rules in ANTLR. "begin" predicts both BEGIN and ID, unfortuantely.

Use the token {...} section:

class TestLexer extends Lexer;
...
tokens {
        BEGIN="begin";
}

ID	:	('a'..'z')+
	;

The lexer will automatically return BEGIN from the lexer if it sees "begin". The lexer maintains a literals hashtable with all the keywords and other strings and then tests for literals before returning to the parser.