It is well known that in Daegu Korean there is a system of lexical pitch accents. However, it has been not clear how the surface tone patterns are in a fall after the H tone and a rise before the H tone. We assumed two hypotheses on this issue. One, called 3-target hypothesis, presupposes an L tone immediately before/after the H tone. The other, called 2-target hypothesis, presupposes no tone between the H tone and the L boundary tone. We conducted an acoustic experiment examining these hypotheses by extending prosodic words through suffixation and compounding. The results for both fall and rise were inconsistent but those for the fall are interpretable as supporting the 3-target hypothesis.