Thread:The S/@comment-995426-20181204095605/@comment-995426-20181206014812

Well, most Japanese long vowels are part of the same morpheme. When the vowel components are split between morphemes, you transcribe them separately in Hepburn romanization. Also, in native and Sino-Japanese words, えい is always ei (not *ē) and いい  is always ii (not *ī).  For instance, the place name 新潟 is always romanized , not *Nīgata, even though the kanji are analyzed as nii-gata and ii is definitely a single long vowel. This leaves ā, ō, ū, and ā already rarely occurs, mainly just in a few words like お母さん okāsan which derived as a phonetic alteration of earlier okassan. This further leaves just ō (for intramorphemic おお o-o and おう o-u) and ū (for intramorphemic うう u-u). But in katakana words, all five long vowels can be transcribed ā, ē, ī, ō, ū if appropriate.