The Complementary Distribution Revisited: nē and nōn in wish and result clauses

For Classical Latin, it is established that the two sentential negators nē and nōn are in complementary distribution (see for example Fruyt 2008a: 8; Pinkster 1986: 147) and for the most part, they indeed appear in separate environments. However, there are two types of clauses, i.e. wish (1a-b) and result (1c-d) clauses, which seem to allow for a free alternation of the negators. This is problematic because this would indicate a violation of the distribution in its strictest sense. 

(1) a. illud utinam ne ue<re>

 that would that NEG true

scriberem

write.SUBJ.IMPF.1SG

(Cic. Fam. V. 17. 32)

b. […] utinam non nocuisset

 would that NEG harm.SUBJ.PLUP.3SG

 (Ovid. Ep. Pont. I. 5. 27)

c. …] ita ut ne altitudine

so so that NEG height

 escendat […]

 go up.SUBJ.PRES.3SG

(Var. R. Rust. III. 27. 1)

d. […] ut amusium non

so that amusium NEG

 desideretur

 want.PASS.SUBJ.IMPF.3SG

(Vitr. Arch. I. 6. 2)

I argue, however, that there is a semantic distinction between nē and nōn which authorises the usage of both in the same environment and also preserves and reaffirms the complementary distribution that has been put forward. Following Mellet (1992), I propose that the choice of negator actually depends on the presence or absence of epistemic modality. She suggests that in non-overlapping environments, nē and nōn can be differentiated on the basis of their interaction with the mood and the modality of the verb. According to her, the subjunctive mood instigates both the possibility of alternative realities and ‘une visée du sujet’ which I reinterpret as the expression of the writer’s (un)certainty towards the given proposition. The choice of negator is then related to both aspects: nē is the negator which does not exclude the alternatives and expresses uncertainty due to its conformity to the subjunctive mood, whereas nōn does cancel them out and expresses certainty due to its assertive nature. In my analysis, I adopted some of Mellet’s concepts in order to find out if such a distinction could also be discovered in the overlapping environments in (1a-b) and (1c-d). For result clauses, the ideas at hand did the trick: All the clauses featuring nē expressed uncertainty about NEG-p, maintaining p as a possible alternative, and all the clauses featuring nōn did the opposite. For wish clauses, however, I suggested an adjustment: The aspect of wishing causes the alterity to shift from p tot ¬p, with ¬p being what is wished for, and thus not realised. The recurrence of the pluperfect tense in clauses featuring nōn indicated that p was already a fact, expressing certainty about the impossibility of ¬p and rendering the wish futile; the recurrence of the imperfect tense in nē clauses made p more undecided, hence turning ¬p into an alternative that cannot be excluded. This analysis demonstrates that nē and nōn are not interchangeable and that despite their alternating occurrence in result and wish clauses, they are in fact in complementary distribution.