Using Monkey Tester app (or a simple app that contains an editable TextArea or TextField), insert the following string:

"Bhojpuri 𑂦𑂷𑂔𑂣𑂳𑂩𑂲 test"

or using the unicode escapes

"Bhojpuri \ud804\udca6\ud804\udcb7\ud804\udc94\ud804\udca3\ud804\udcb3\ud804\udca9\ud804\udcb2 test"

- position the cursor at the end of text
- press Option+LEFT twice

Observed: the cursor first jumps to the beginning of the word 'test' then to the beginning of the word 'Bhojpuri'
Expected: the cursor should be positioned at the beginning of the word '𑂦𑂷𑂔𑂣𑂳𑂩𑂲'

The culprit is the code in TextInputControl:814 which calls Character.isLetterOrDigit(char) on a character which is a part of a surrogate pair instead of Character.isLetterOrDigit(int) on the code point.

The 'next word' function works correctly.

The issue is observed on macOS 14.4.1.

Added unit test in TextInputControlTest:

    @Test
    public void previousWord_Bhojpuri() {
        // "Bhojpuri \ud804\udca6\ud804\udcb7\ud804\udc94\ud804\udca3\ud804\udcb3\ud804\udca9\ud804\udcb2 test"
        textInput.setText("Bhojpuri 𑂦𑂷𑂔𑂣𑂳𑂩𑂲 test");
        textInput.end();
        verifyCaret(28);
        textInput.previousWord(); // at the beginning of "test"
        verifyCaret(24);
        textInput.previousWord(); // at the beginning of "𑂦𑂷𑂔𑂣𑂳𑂩𑂲"
        verifyCaret(9);
        textInput.previousWord(); // at the beginning of "Bhojpuri"
        verifyCaret(0);
    }