In this experiment, one digit arithmetical problems followed by a masking sound were given auditorily from a computer. Three types of calculation tasks; addition tasks, multiplication tasks, and kuku tasks, were tested under two conditions that varied depending on the position of the masked sound in a given formula. For example, when a left side of a calculating formula consisted of number 6 and 7, the addition task was conducted as “6 + X (masking sound) = 13 (roku, tasu, X, wa, juusan)”, or as “X + 7 = 13”, and the multiplication task was presented as “6 × X = 42 (roku, kakeru, X, wa, yonjuuni)”, or “X × 7 = 42”, while the kuku task was presented as “roku, X, shijuuni”, or as “X, shichi, shijuuni”. When each stimulus was presented, each of the participants of 10 men and 10 women was required to respond by answering with what was missing. The results revealed that they answered faster in the kuku tasks than in addition tasks. The results indicate the possibility that calculation by kuku was mostly executed through a process similar to a playback of verbal memory stored as linguistic representation, and when we could ascertain the kuku tasks, the quantitative representation of numbers almost do not come to the surface. Thus, Japanese adults answered faster in the kuku tasks than in the addition tasks.