bind() now sets necessary attributes automatically (autocomplete, spellcheck etc)
toKana() no longer converts zenkaku latin toKana now: ('ｉｍｉ imi' => 'ｉｍｉいみ')
toRomaji() extends vowels for katakana chōonpu, IE: ゲーム => geemu, toHiragana() converts to hyphen => ge-mu
toKana() without IME mode converts lone 'n' => 'ん', 'nn' => 'んん' if no other chars present
isJapanese() now returns false for latin numbers (201) (still true for zenkaku ２０１)
tokenize() splits into finer categories, view API documentation and tests for full details
rewrite of conversion methods to allow custom mapping adapters

isJapanese and isRomaji now accept a second param, a regexp that will also pass the check
global option romanization for toRomaji() (currently only 'hepburn' however)
global option customKanaMapping for toKana()
global option customRomajiMapping for toRomaji()

inserting text between already existing characters in a bind() IME input field now properly sets cursor to correct position after conversion
toRomaji() little ヶヵ used in words like 一ヶ月 are no longer converted since they are used as symbols (like the kanji) and do not actually denote kana. Previous behaviour: 一ヶ月 => 一 ke 月
toRomaji() no longer incorrectly duplicates non-glottal stops following っ. Previous behaviour: あっ、 -> a,,