This paper examines tasi and tto, two Korean adverbs with roughly the same semantic value of again. It presents two empirical generalizations: (i) semantically tasi can mark the recovery of the original state, modifying an eventuality that is token-identical to a contextually-given eventuality, whereas tto is only limited to mark a simple repetition, thus modifying a type-identical event; (ii) functionally tasi can modify result states, but tto cannot. Then, it illustrates how the functional difference of result state modification can be accounted for by the notion of default events.