Today, when the birds were whistling in our garden, the sun was playing hide and seek with the clouds that were driven by the engine of the wind… the question: “how can we recognize a null character” popped up again.
After doing some deciphering and reading, on my desk were the prints of all the occurrences & dependencies of all words and letters in them of the VMS that I’ve made before.
This time I wrote down in columns the high occurrences of letters left (L1) and right (R1) and then realized that
- when a letter on Apos of a word (initial word pos.) has almost the same characteristics as that same letter inside a word then, that letter is safe.
- All the letters that do not conform to that simple rule, are very candidates for a null, because they behave differently on different places.
- When a letter is unpredictable on different places and, yet at the same time has at least a mediocre occurrence rating, it has become a serious candidate for a null.
- If the candidate also shows up on almost any letter left and right, and it’s on any position in any word, then it’s a strong null candidate.
Be aware that a standard vowel can do such tricks in some languages as well.