Hebrew has five letters whose shape depends on position: they take a special final (sofit) form at the end of a word and a regular form everywhere else. This tool detects and corrects letters written in the wrong form for their position.
How it works
The normaliser scans each letter, decides whether it stands at the end of a word, and applies the placement rule:
final forms: ך ם ן ף ץ (kaf mem nun pe tsadi)
regular forms: כ מ נ פ צ
rule: letter at word end -> final form
letter not at word end -> regular form
word boundary = any non-Hebrew-letter character (space, punctuation, digit)
A letter is “at word end” when the next base Hebrew letter does not exist or a non-letter follows. Niqqud (vowel points) and cantillation marks are counted as belonging to the current word, so a vocalised word-final letter is still recognised correctly.
Tips and example
שלומ is corrected to שלום because a word-final mem must be the sofit form ם, and a stray ך inside a word would be changed back to כ. This is the typical clean-up needed when text has been typed without final-form awareness or copied from a source that lost the distinction. Review the change list before copying, since proper nouns and acronyms occasionally use forms deliberately.