Lookup Tables and the Language of Clicks

The telegraph sounder on my bench produces two distinct sounds: a sharp click when the electromagnet pulls the armature down, and a softer clack when the spring releases it back. What matters isn’t the individual sounds but the time between them — short gap means dot, long gap means dash. And once you have a sequence of dots and dashes, you need a way to look up what letter they represent.

This is a symbol table: a structure that maps keys to values. In Scheme, the classic approach is an association list — pairs of (key . value) that you search with assoc:

(define morse-table
  '((".-" . #\A) ("-..." . #\B) ("-.-." . #\C)
    ("-.." . #\D) ("." . #\E) ("..-." . #\F)
    ("--." . #\G) ("...." . #\H) (".." . #\I)))

(define (decode-letter code)
  (let ((pair (assoc code morse-table)))
    (if pair (cdr pair) #\?)))

(map decode-letter '("...." ".." "-.-." "." "-.." ".-"))
;; => (#\H #\I #\C #\E #\D #\A)

Java prefers explicit hash maps, which trade the elegant simplicity for O(1) average lookup:

import java.util.*;

public class Morse {
    static Map<String, Character> table = new HashMap<>(Map.of(
        ".-", 'A', "-...", 'B', "-.-.", 'C', "-..", 'D',
        ".", 'E', "..-.", 'F', "--.", 'G', "....", 'H', "..", 'I'));

    public static void main(String[] args) {
        for (String code : ".... .. -.-. . -.. .-".split(" "))
            System.out.print(table.getOrDefault(code, '?'));
    }
}
// Output: HICEDA

Both implementations do the same thing: accept a sequence of dots and dashes, return a character. The Scheme version walks the list sequentially — O(n) per lookup — while Java hashes the key to jump directly to the bucket. For the 40-ish codes in International Morse, the performance difference is negligible. For a vocabulary of millions, it would matter.

What I find interesting is that the telegraph operators of the 1860s were the lookup table. They heard the clicks, mentally matched the pattern, and transcribed the letter. The table lived in their heads, trained through repetition until it became automatic. My thirty years of ham radio gave me that same mental table for International Morse — dah dah dah is O, obviously — but the sounder on my bench was built for American railroad Morse, where O is di-space-di. The codes are different. My mental table is wrong for this machine.