Test takers’ score development of a computer based English speaking test is affected by task types and scoring constructs and yet there is little literature on the relationship between them. This study investigated the effects of task types and scoring constructs on test takers’ speaking performance over a period of one year. For the purpose of this study, 46 Korean university students were given an English speaking test three times. Based on ACTFL scoring guidelines, six task types were proposed: self-introduction, read-aloud, personal questions, picture and graph description, and giving opinions. The six scoring constructs of the speaking test were task completion, coherence, fluency, pronunciation, language use and grammar. The results revealed that the means of both task types and scoring constructs were significantly differently increased over the period. Secondly, the results indicated that the read-aloud task was the easiest, while the giving opinions task was the most difficult. Lastly, the score of task completion was easily improved while the scores of accuracy factors such as grammar and language use needed more time for improvement. Some suggestions are discussed in terms of task types and speaking performance.