This article describes the prosodic characteristics of utterances with a broad focus, narrow informative focus and narrow corrective contrastive focus in Apartado and Medellin, Colombia. A corpus of 195 utterances elicited from semi -spontaneous speech and sentence reading is analyzed. The prosodic analysis is based on the metric-autosegmental model, with its Sp_ToBI labeling system and the behavior of declination, speed, tonal range, intensity, and duration. The results suggest a relevant influence of the acoustic parameters and some social variables of the collaborators in the performance of the different types of focus.