Consider the definition of tt(input): verb( input: // empty | input line ; ) This definition should be read as follows: em(A complete input is either an empty string, or a complete input followed by an input line). Notice that `complete input' is defined in terms of itself. This definition is said to be em(left recursive) since the rule's nonterminal (tt(input)) appears as the leftmost symbol in the production rule. See section ref(RECURSIVE). The first alternative is empty: there are no symbols between the colon and the first `tt(|)'; this means that input can match an empty string of input (no tokens). We write the rules this way because it is legitimate to type tt(Ctrl-d) (end of input) immediately after starting the calculator. By convention empty alternatives are provided with the comment `tt(// empty)'. The second production rule (`tt(input line)') handles all nontrivial input. It means em(after reading any number of lines, read one more line). The left recursion turns this rule into a loop: it processes any number of lines. The parser's parsing function continues to process input until a grammatical error is encountered or untilthe lexical analyzer returns the end-of-file token.