If the previous character is a space character and nWordType indicates spaces should be skipped, and if the first non-space character is an Asian character, then, since Asian word break needs language specific wordbreak dictionaries, the method will return -1 in Boundary::endPos and the position after the Asian character (i.e. the space character) in Boundary::startPos. The caller then has to call this method again with a correct aLocale referring to the Asian character, which is then the previous character of the space character where nStartPos points to.

Note that the OpenOffice.org 1.0 / StarOffice 6.0 / StarSuite 6.0 i18n framework doesn't behave like this and mixed Western/CJK text may lead to wrong word iteration. This is fixed in later versions.

If nPos is the boundary of a word, it is StartPos of one word and EndPos of previous word. In this situation, the outcome of the algorithm can be indeterminate. In this situation the bPreferForward flag is used. If bPreferForward == false, nPos is considered to be the end of the word and we look backwards for beginning of word, otherwise nPos is considered to be the start of the next word and we look forwards for the end of the word.

This method is mis-defined, since WordType is not an attribute of a word, but a way to break words, like excluding or including tail spaces for spellchecker or cursor traveling. It returns 0 always.

It is possible that both of this method and following method isEndWord all return true, since StartPos of a word is inclusive while EndPos of a word is exclusive.