Processing Negation in a Miniature Artificial Language

In two miniature artificial language learning experiments, we compare the processing of narrow and broad negation, corresponding to prefixal negation (unhappy) and free-standing negation (not happy) respectively, with that of non-negation (happy). Three artificial prefixes were invented to express the three meanings above. The meaning scope expressed by the negation types was manipulated in the experiments, and the processing of the three forms was tested through a picture-word verification task. In Experiment 1, the scope expressed by prefixal negation was included in the scope expressed by free-standing negation, while in Experiment 2, there was no overlap between the two negation types and the scope of free-standing negation was limited to the intermediate range of a scale. Experiment 1 showed that narrow negation is more difficult to process than the non-negated meanings, but not as difficult as broad negation. Experiment 2 showed that when the meaning scope of broad negation was restricted to the middle range, the processing difficulty found in Experiment 1 disappeared, as it did not take longer for participants to identify the middle range compared to the ends of the scale. We show that the chunking of the negated meanings relative to one another plays a role in the processing cost of these forms.