In this paper I investigate some empirical and theoretical issues regarding the presence of negation in imperatives, focussing on Korean. Korean has a special form of negation which appears in imperative clauses, and I argue that the conditions governing the choice of regular negation or imperative negation are purely semantic in nature. In particular, I argue that imperative negation constrains the denotation of its clause to be of type OUTCOME in the system of Ginzburg and Sag (2000), as opposed to regular negation which appears with PROPOSITIONS or QUESTIONS. OUTCOMEs are built from irrealis soas (i-soa) while the others are built from realis soas (r-soa). The partitioning of the two negations must go along these high-level semantic lines. As in other languages with polite or honorific morphology, clause-type distinctions that Korean makes morphologically are typically neutralized in the presence of honorific morphology, and in such cases, using one negation or other disambiguates an example to just those clause denotations built out of an i-soa or an r-soa. I also compare a lexically-based HPSG analysis with a derivational account treating imperative negation as the spell-out of a combination of abstract functional heads.

• short version of the paper
• The bibliographic reference:

  @inproceedings{Sells2003,
  author    = {Peter Sells},
  title     = {Negatives of Imperatives},
  booktitle = {Proceedings of the 10th Harvard International Symposium on Korean Linguistics},
  publisher = {Harvard University},
  year = 2003