Jaume Solà i Pujols - Departament de Filologia Catalana ...
Jaume Solà i Pujols - Departament de Filologia Catalana ...
Jaume Solà i Pujols - Departament de Filologia Catalana ...
Create successful ePaper yourself
Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.
ecause control would be a universally available grammatical option, which 'tries to apply'<br />
whenever possible. This i<strong>de</strong>a could be related to the Elsewhere Condition, which essentially<br />
states that regular processes apply automatically unless they are blocked by the existence of more<br />
specific/irregular processes (e.g., in morphology, a regular verbal form is used unless an irregular<br />
form exists).<br />
Suppose control is a universal option, in fact the most 'regular' one possible for<br />
infinitives, and applies whenever possible, unless a more specific/idiosyncratic option exists (for<br />
instance, an idiosyncratic subcategorization specification for 'believe' ([__IP]) in English). Then<br />
the existence of AGR in lexical control structures is forced by the obligatory application of<br />
control: control involves PRO; PRO can only appear in Spec of AGR; 139 an element in Spec of<br />
AGR has only two options to be licenced, in accordance with our previous assumptions: either it<br />
becomes an AGR-i<strong>de</strong>ntifier or it is resumed by an in<strong>de</strong>pen<strong>de</strong>ntly Case-marked I-subject (see next<br />
chapter). Putting asi<strong>de</strong> the second option, the conjunction of the above assumptions gives the<br />
result that AGR has to be present in control structures. 140<br />
139 It is far from obvious that theories being as rich in<br />
structural positions as the Split-INFL hypothesis, or Larson's<br />
(1988) VP-shell theory, can manage to confine PRO to Spec of<br />
AGR. Two solutions come to mind: a) PRO (or PRO o ) is inherently<br />
an AGR-i<strong>de</strong>ntifier; b) PRO can appear in other positions, but<br />
then it is not available for control.<br />
140 It is not obvious that the Elsewhere Condition, which has<br />
been used in phonology and morphology, should be relevant for<br />
syntax. I think it is at variance with Chomsky's (1988)<br />
implementation of the i<strong>de</strong>a of economy of <strong>de</strong>rivation, in that he<br />
explicitly states that regular/universal options take preference<br />
over irregular/idiosyncratic options. In fact Chomsky's proposal<br />
is inten<strong>de</strong>d to <strong>de</strong>al with subtle, theory internal problems, while<br />
the EC could be argued to be relevant for syntax in more obvious<br />
cases (e.g., if a language has object agreement morphology, it<br />
has to use it, and give up what is likely to be the more<br />
universal option of not using it). In addition, the i<strong>de</strong>a of<br />
economy need not be tied to Chomsky's universal-over-particular<br />
constraint. For instance, Roberts (1991-a) uses a notion of<br />
economy based on the length of <strong>de</strong>rivations.<br />
Since, however, Chomsky is concerned with the economy of<br />
<strong>de</strong>rivations in a very subtle sense, it could turn out not to be<br />
at variance with the Elsewhere Condition if some subtle<br />
distinction could be ma<strong>de</strong> between the field of application of<br />
1