Abstract
Negative Concord (NC) is analysed as a syntax-morphology interface phenomenon. It can be classified as Strict or Non-Strict depending on whether n-words necessarily co-occur with the sentential negative marker (SNM) in all contexts or not. In Non-Strict NC, only post-verbal n-words must co-occur with the SNM; pre-verbal n-words cannot occur with the SNM without yielding Double Negation.Adopting the Minimalist Program and the Distributed Morphology model, it is argued that two post-syntactic operations determine the distribution of n-words with respect to the SNM both in Standard and Non-Standard English, and in Romance. Standard English n-words such as nobody, nothing and the like are assumed to be non-negative indefinites.
Standard English is analysed as having the structure of an NC language. However, Obliteration, a PF operation triggered by a language-particular Filter that disallows the accidental repetition of negative features in the same Spell-Out domain, prevents n-indefinites from co-occurring with the SNM.
When Obliteration cannot rescue the violation of the Filter, Impoverishment applies. It deletes the (uninterpretable) negative feature of an n-indefinite and results in the insertion of a default form of the any-set.
Phase Theory determines whether an n-word and the SNM are in the same Spell-Out domain. Assuming the complement of a phase head to be sent to Transfer at the end of a phase, object n-indefinites that raise to Spec, NegP, as well as subject and fronted n-indefinites occur in the same Spell-Out domain as the SNM. Conversely, object n-indefinites that remain VP-internal, are not in the same Spell-Out domain as the SNM. While in Standard English n-words obligatorily raise to Spec, NegP, they do not have to in Non-Standard English. This explains why post-verbal n-indefinites can co-occur with the SNM in both Strict and Non-Strict NC varieties of Non-Standard English. Data from the FRED corpus show that Obliteration is also attested Non-Strict NC varieties of British English as, like Standard English, these are constrained by the Filter that prevents two negative features from being accidentally repeated in the same Spell-Out domain.
The conclusions for Non-Standard British English can be extended to African American English (AAE), which mostly implements a system of Strict NC. For some AAE speakers, though, NC is Non-Strict.
Such variability in the NC-type makes the study of negative inversion (Nl) in AAE extremely interesting as Obliteration is unexpectedly suspended in Nl constructions even for those speakers with a system of Non-Strict NC. I claim that Focus is involved in Nl, which explains why the negated auxiliary precedes the subject. Obliteration is argued to be blocked on semantic grounds, as it would result in a syntactic structure that would be interpreted as a question if the SNM were Obliterated.
For Non-Strict NC Romance languages (e.g. Italian and Spanish), Obliteration applies with pre-verbal n-words, in being in the same Spell-Out domain as the SNM. Post-verbal n-words, conversely, co-occur with the SNM because negation is merged on top of TP. It is further argued that Romance n-words are underspecified for polarity, which allows them to occur in a wide range of contexts, including negative constructions. Standard French bears striking similarities with Standard English. The SNM pas cannot co-occur with n-indefinites in any contexts and is affected by Obliteration. Like in Standard English, negation in French is assumed to be merged on top of v*P. Raising of n-words to Spec, NegP results in Obliteration both with pre- and post-verbal n-words.
Finally, Strict NC Romance languages (e.g. Romanian) do not display the effects of PF-operations because they are not sensitive to the Filter that prevents two negative features from accidentally occurring in the same Spell-Out domain.
Date of Award | 2 Jun 2008 |
---|---|
Original language | English |
Supervisor | Mireia Llinas Grau (Director) & Josep Quer Villanueva (Director) |