There is a substantial discrepancy between normative and empirical views on Dutch standard pronunciation. The discrepancy between these views can be reduced by looking at the empirical range as being structured from an imagined point of reference. The prescriptive standard then operates as the ultimate model toward which the submodels for standard pronunciation are oriented. A study investigated whether segmental deviations from Dutch standard pronunciation affect listeners' evaluative reactions to speakers according to the nature or quality of those deviations. A 30-second fragment of neutral prose was read by seven speakers, four Dutch and three Flemish, two characterized as standard speakers and five with regional accents. From each fragment, 15 words (identical for each speaker) representing 25% of the fragment were isolated. The words contained 27 kinds of deviation from standard pronunciation. The words were presented to about 50 students, who were asked their perceptions of the speech and of the speakers. Some segmental features that may have played a role in differential attitudes toward varieties of Dutch were indicated. Phonological deviations from standard Dutch appeared to be more important in language attitudes and to get a more extreme evaluation, positive or negative, than phonetic deviations. (MSE)