I do not do English lyrics but try to tweak Synth V to pronounce German instead. However, I think I get better results by following a C-V (or C-V-C) syllable logic, which should match the Japanese language structure more closely. I figure the whole engine might be built upon this.
So (p r ao) (p ow) (z ih) (sh ax n uh)
I always add a short extra vocal at the end (duration 20 to 40 %), uh is usually fine, to get a voiced vocal, which in Asian languages would certainly be regarded as a new syllable.