Abstract We describe an experiment based on introspective assessment of inter-syllabic prosodic changes in two different conditions (global - continuous sentences, and local - two-syllable portions of speech), by means of discrete targets (Higher, Same and Lower). The results show that global evaluation is very problematic for the listeners and that f0 is the best assessed parameter, followed by intensity and duration. They also provide a realistic, continuous-speech-based view of just noticeable differences, at least for f0 and intensity.