I cannot reproduce this issue, sorry.
When I have researched long form subtitles, the recommendation has often been to use an Illustrator or Photoshop file instead of dozens or hundreds of individual text effects. The transcript is pasted into a canvas that is maybe 6,000 pixels (or more) tall by 1920 pixels wide. Formatting and styles are applied to the graphics file, use double or triple paragraph spacing between the individual groups of text so there's a visual gap, and save it with transparency.
Import the file into FCPX ad drop it onto the timeline. Make sure the alpha is properly interpreted. Adjust the vertical position so the first lines of text appear Apply masking if necessary. You apply a hold-style keyframe for every time the text needs to advance. If you have formatted th text prp[erly, the vertical change in position is always the same.
The wisdom of this approach becomes clear if you need to make any changes to the text file. As long as the paragraph formatting doesn't change, you never need to reprogram any of the vertical keyframes.
There are lots of ways to do subtitles including setting them up as captions that can be turned on or off in most players instead of burning them permanently into the video. And there are subtitle plugins (try FxFactory) that provide far more control and flexibility, as well as some text file importing capabilities,
