Most of the evidence for the role of BA 44 in syntactic processes, be it noted,
comes from experiments using invented lexical items or artificial grammar (e.g.,
Tettamanti et al. 2002, 2009, Friederici et al. 2006, Bahlmann et al. 2008).^4
BA 44 has also been implicated in non-linguistic tasks. BA 44 is selectively
implicated in the acquisition of “non-rigid” dependencies established between
visuospatial symbols (Tettamanti et al. 2009), which clearly shows that this
portion of the brain is responsible for processing hierarchy in general. Process-
ing of hierarchical object manipulation is also reported to activate BA 44, which
has been proposed as the neural correlates of Greenfield’s (1991) “Act ion
Grammar” (Fazio et al. 2009, Higuch i et al. 2009).^5 A virtual lesion of BA 44
causes a problem with hierarchically encoding other people’s behavior (Clerget
et al. 2009). Base d on a large body of literature on BA 44, Yusa (2012a, b)
suggests t hat domain-ge neral Merge might be implicated in BA 44, which
modulates the processing of hierarchical structure across cognitive domains.
Given this suggestion, it is not surprising that BA 44 is activated in both lin-
guistic and non-linguistic tasks. It is therefore not the case that hierarchy should
be a problem with language evolution, as Frank et al. (2012) claims.
On the other hand, evidence for the role of BA 45 in syntactic processes
comes from the acquisition/processing of real rules in natural languages (Musso
et al. 2003, Yusa et al. 2011).^6 BA 45 is also responsible for embedding of
syntactic constituents (Shetreet et al. 2009). The number of embeddings affects
BA 45 more than BA 44 (Pallier et al. 2011). BA 4 5 is also reported to be
selectively activated by movement or internal-Merge operations (Santi and
Grodzinsky 2010). If internal-Merge is a unique property of human language,
it is not implausible to assume that BA 45 might have something to do with
deep properties of human language. Based on these observations, Yusa (2012a,
b)suggests t hat domain-specific Merge might be implemented in the pars
triangularis. Recently, Fedorenko et al. (2012) cl aimed that traditional group-
based fMRI analyses have obscured the “functional heterogeneity within Broca’s
area.” Individual-subject fMRI analyses clearly show that the pars triangularis
is robustly sensitive to linguistic stimuli but not to non-linguistic stimuli. This
is in conformity to the claim that BA 45 is responsible for domain-specific
Merge (see Yusa (2012b) for more discu ssions of BA 44 and 45 and Fujita
and Yusa (submitted) for the evolution of language in terms of Merge in
the brain).
With regard to BA 44, a dorsal pathway connecting BA 44 and the posterior
temporal lobe via the arcuate fasciculus is hypothesized to be involved in map-
ping sound to articulation. In contrast, a ventral pathway connecting BA 45
and the temporal lobe via the extreme capsule is responsible for the mapping
of sound to meaning (Hickok and People 2004, Weiller et al. 2009).^7 Indeed,
the functional roles of these pathways continue to be discussed, but it is worth
bearing in mind that a better understanding of the essential properties of these
pathways converging in Broca’s area might provide new insights into how unique
language is to humans.
Syntax in the brain 225