| description |
In the Minimalist Program (1995), Chomsky gets rid of the Agreement projection and argues that Tense and Agreement can be assimilated universally. Tasks that were previously thought to be responsibility of the Agreement node are reassigned to Tense and other heads. Serbo-Croat data contradicts the above theory showing that not only does this language lack a Tense projection, but it also lacks a Tense marker, allowing a manipulation and interpretation of its predicate forms. Further investigation shows that the entire Slavic language group, as well as some African and some Arabic languages display the same property. A Tense morpheme is absent from all of them and they heavily rely on Aspect in the computation of the time of any given action. Apart from Slavic languages, a link between Tense and Aspect is evident in Spanish and Latin as well. Aspectual opposition is not powerful enough to express all the various "time frames" that a language needs and that are, for example, available in English. To overcome this problem, Slavic languages introduce Agreement into the "time" computation. Introducing two different sets of Person and Number markers ensures even greater variety. Since the Tense head is absent from Serbo-Croat, Nominative Case assignment is performed by Agreement. This phenomenon is also characteristic for Bulgarian (and possibly the entire Slavic group, as well as some unrelated languages, like Portuguese and Galician). Consequently, Serbo-Croat does not allow Exceptional Case Marking but instead inflects its infinitives for Person and Number. Surprisingly, the absence of a Tense marker does not imply the absence of the [Tense] feature as all Tense less languages seem to be perfectly capable of communicating the time and duration of any given action. Thus, I have also found that the [Tense] feature does not have to be morphologically realised, although all languages invariably seem to display Aspect morphology. To account for the properties of both Tenseless and Tensed languages, I provide the evidence for the argument that languages universally project Infl and that both [Asp] and [T] features count as its intrinsic categorial features, in order to survive until LF. We also know that this projection must be specified for a strong [N] feature that forces overt subject raising, and that it is not specified for a strong [V] feature, as the main verb does not leave the VP until LF. In languages that do have a Tense marker, like in English, the [Tense] feature is checked at LF by the Tense marker. In languages that do not have a Tense marker, like Serbo-Croat, the [Tense] feature gets checked at LF in an alternative way. A number of smaller but related issues are also handled on the way, particularly PRO distribution. I argues that PRO is assigned null Case if and only if it is controlled (by subject, object or arbitrarily) and that PRO acquires the null Case from its controller, not from the Infinitive. Contrary to Boskovic's (1996) proposals, uninflected Infinitives can not assign any Case in any language.
|
