3 Ordering: third-factor underspecification
Third factors are “principles not specific to the faculty of language” according
to Chomsky’s (2005) defi nition. In other words, principles belonging to the
third factor are obeyed not only within the language system but also in other
biological, physical or computational systems. This means that explanation by
appeal to the third factor can be regarded as having greater generality than one
based on the first factor, in the sense that any appeal to UG, whether parameter-
ized or not, is circumvented under third-factor explanation thereby contributing
to the goals of (i) getting the facts right, e.g. a frog’s DNA does not in fact
determine the shape of a frog’s cells, rather this frog-property is determined by
physical law) (ii) explaining language evolution in a manner consistent with both
(a) Occam’s razor, and (b) the known facts indicating the existence of a sudden,
recent, and simple mutation. Obata, Epstein and Baptis ta (2015) propose that
variant “rule ordering” can underlie (cross)linguistic variation. How can linguistic
variation be explained in terms of such (by hypothesis) third factor licensed
ordering variation? Obata, Epstein and Baptis ta (2015) demonstrate that variant
orders of applying (universal) syntactic operations can generate multiple gram-
matical outputs, each of which equally satisfies the interface conditions and each
of which is derived in a computationally efficient manner. That is, given the
independently motivated definition of “computationally efficient satisfaction of
the interfaces” – not proposed in order to capture variation – there are cases in
which more than one derivational rule ordering is optimal. In cases in which the
third-factor content of the theory independently entails that more than one rule
ordering is optimal, the possibility of (cross)linguistic variation is predicted. The
question then is whether these predictions, once identified, are correct and, if
any are, how much variation might be explained in these terms? Obata, Epstein
and Baptista (2015) explain variation of C-agreement in Cape Verdean Creole and
Haitian Creole and also agreement variation in English and Kilega by appeal to
different timing/ordering of agreement and Merge. (See Obata, Epstein and
Baptista (2015) for details of the analysis and also Obata and Epstein 2011,
201 2 for relevant analysis. ) We will now present additional cases which can be
similarly captured under ordering variation permitted by third-factor underspecifi-
cation of what constitutes computational efficiency.
Chomsky (2013) discusses Au x-inversion with matrix wh-movement in English
and argues that C searches T by (the third-factor principle of) minimal search,
before the subject moves to T-Spec as in (2).^1
(2) C’s minimal search be fore subject-movement finds TC
T
D/Subj ......
132 Miki Obata and Samuel Epstein