This problem occurs in both Windows NT and Solaris 2.5.1 running under a ja locale.

1. Load and Compile the attached BreakMe.java file.

2. In Solaris 2.5.1 environment run the BreakMe.java application. This application is made of of three elements, a TextField to enter input at the top, a TextArea which displays output in the center, and a button to start parsing at the bottom.

3. Open and copy the text from the attached BreakMe.inputtext (EUC) encoded and paste the text into the upper TextField of the BreakMe.java application.

4. Press the parse button.

5. The application will break the entered text in all areas where line breaks should occur and place the remaining text on the line below.

6. The expected results should follow the Japanese Line Breaking rules known as "kinsoku." For more information cocerning this please either refer to Teruhiko.Kurosaka@Japan. The current BreakIterator.getLineInstance() does prevent the opening Japanese quotation mark from being at the end of a line and the closing quotation mark from being at the begining of a line. However it does not prevent the Japanese comma or period from being at the beginning of a line.