(2003 ). Moreover, contrary to Musso et al. (2003 ), exposure to the relevant L2
input in this study was not brief, providing the participants with situations where
linguistic knowledge was engaged. Our results show that structure dependence is
still operative in the acquisition of a new rule by late second language learners,
who do not learn a rule violating the principle of structure dependence.
It is worth noting here that structure dependence was regarded as pure UG
before the Minimalist Program because there was no serious thought of trying
to extract ‘third factor’ elements from structure dependence (Noam Chomsky, pc).
In the Minimalist Program, however, UG consists only of the simplest recursive
operation called Merge, which applies to two syntactic objects X and Y and
merges them to form a new syntactic object Z: structure principle restricts
computations to order-free structures (with externalization ancillary) and relies
on the simplest computational procedure. The formation of polar interrogatives
involving subject-auxiliary inversion is a classic example of structure dependence.
Chomsky (2013) argues that subject-auxiliary inversion is derived from the
operation of minimal search (as a third-factor principle of computational effi -
ciency) along with the predicate-internal hypothesis. The results reported in
Musso et al. (2003) and Yusa et al. (2011) could be interpreted as providing
neuroimaging support for the operation of minimal structural search in language
processing. In the next section, we will look in more detail at Broca’s area, in
which Merge might be implemented.
5 Broca’s area: BA 44 and BA 45
Broca’s area, usually defi ned as the pars opercularis (BA 44) and the pars trian-
gularis (BA 45) of the left inferior frontal gyrus (IFG), has been reported to be
involved in linguistic and non-linguistic process (Tettamanti and Weniger 2006,
Friederici et al. 2006). Note here that the posterior part (BA 44) of Broca’s area
extends into the premotor area (BA 6) and the anterior part borders the pars
orbitalis (BA 47). Syntactic processes have been reported to be implicated in BA
44 or BA 45,^3 though these two parts are cytoarchitectonically heterogeneous:
BA 45 has a granular layer IV and BA 44 is dysgranular. The agranular frontal
operculum (Friederici et al. 2006) and premotor cortex (BA 6) (Friederici 2006)
are active in the processing of fi nite state grammar. In terms of the principle of
gradation (Sanides 1962), “the brai n evolution proceeded from agranular to
dysgranular and then to complete granular cortex” (Friederici 2011: 186). I n
this sense, the dysgranual BA 44 is unique in that it occupies a place between
the phylogenetically older agranular BA 6 and the phylogenetically younger
granular BA 45. This uniqueness of BA 44 has interesting implications for lan-
guage evolution (Yusa 2012b). It is not i mplausible to stipulate that the functions
in BA 45 are evolved from those in BA 44, which are in turn evolved from those
in BA 6 (See Fujita 2014 for a discus sion of the evolution of Merge). Given
this, it is not implausible, either that hierarchy in general is processed in BA 44
and linguistic hierarchy is in BA 45, which will be discussed below.
224 Noriaki Yusa