==Kana tanpa bentuk romaji==
There is no generally accepted form of romanization for some forms of kana. In particular there is no form of romanization for full-sized kana combined with smaller versions of the vowel kana, "ぁ", "ぃ", "ぅ", "ぇ" and "ぉ", the smaller versions of the ''y'' kana, "ゃ", "ゅ", and "ょ", and the [[sokuon]] or small ''tsu'' kana "っ". Although these are usually regarded as merely phonetic marks or [[diacritic]]s, they do appear on their own, for example at the end of sentences or in some names.
There is also no commonly accepted way of romanizing common combinations such as "トゥ" of katakana ''to'' and small ''u'', used to represent sounds as in the English word "too". Some people write this pair as ''tu'', but this is likely to be confused with the ''tu'' Nihon-shiki and Kunrei-shiki romanizations of the kana ツ, romanized as ''tsu'' in Hepburn romanization.
On a computer or word processor, these smaller kana may be produced in various ways. For example, an "x" or an "l" preceding the romanization of the full-sized kana produces a small version on some systems, thus ''xtu'' gives "っ" on a Microsoft computer. However this is not standardized, and these forms are restricted to use in [[Japanese input methods|input systems]]; they are not used to represent the smaller kana in romanized Japanese.