This change makes sure that empty strings are sorted before non-empty strings. Note that it is treating the case where both strings are equal specially, this is for consistency with the logic at the bottom of compareTextsAlphabetical() and compareTextsCharacterCode(). I don't see a difference if I leave out this case but if it is important then personally I would have written this differently:

if (aS2.length() == 0) {
    return false;
} else if (aS1.length() == 0) {
    return true;
}

The checks in compareTextsAlphabetical() can be simplified similarly:

if (lPositionS2 == aS2.length()) return false;
else if (lPositionS1 == aS1.length()) return true;

These two comparisons produce the same result as the six comparisons currently there.

See STR in bug.