What the previous two posters have failed to mention (despite being tutors...?) is that a speaking/shouting tone of voice relies heavily on pitch variation. For reference you may want to record yourself speaking those parts of the song so that you can easily check how the pitch changes, or even open it in a pitch analysis viewer (like vocalshifter or any other pitch correction software with a visualizer). Then you can replicate it in UTAU.
If the lines are rhythmical and match to the beat of the song, you're in luck because you won't have to worry about the timing as much. (With hanasu/talkloid skits, the timing is another major factor in getting the right intonation, and it's tricky because UTAU works on musical timing.)
For a single shout try a descending bend like this:
For a longer spoken phrase, it depends on the language, but for English I would recommend having stressed syllables with higher pitch than other syllables. In general you would want the pitchbends to sort of slur through all the notes as a continuous change, rather than having the notes at a particular pitch.
And keep in mind that singing tends to use a higher range than the same voice speaking! So make sure that you have your speaking parts at a lower range- a common issue in talkloids is when they speak in a really high voice, since that's the range they use for singing.