which stand lower in the scale of psychological evolution and development.” Later
to be known as Morgan’s Canon, this criterion has served as a good antidote
for excessive assimilation or identification of human and non-human cognition.
A fairly recent example of violating this canon may be found in the so-called
Integration Hypothesis of human language evolution (Miyagawa et al. 2014).
This hypothesis maintains that the non-finite state nature of human language
results from the combination of two separate systems, E(xpressive) and L(exical),
each being finite-state in isolation and found in other species (E in songbirds
and L in monkeys, etc.). The conceptual advantage of this hypothesis is that it
allows us to understand language evolution in perfect conformity to the general
pattern of evolution as creation of something new by combining old things.
The empirical disadvantage is that birds’ E-system and monkeys’ L-system, to
the extent that they exist, differ greatly from the human counterparts (roughly,
the systems of functional categories and lexical categories). In particular, the
human system of lexical categories is extremely rich and generative and always
mind-dependent when making reference to the outside world. As Chomsky
has stressed on many occasions, human language has no direct reference rela-
tion, in sharp contrast to animal communication systems, which are related to
mind-independent entities (Chomsky 2007). This gap we really need to mind.
5 Fallacy of the FLN/FLB dichotomy
The above discussion is directly connected to the reevaluation and rejection
of Hauser et al.’s FLN/FLB dichotomy. This dichotomy is based on the sup-
position that many but not all aspects of language are neither species-specifi c
nor domain-specifi c. But exactly in what sense are they not species-specifi c or
domain-specifi c?
Take the SM system as an example of the non-FLN part of FLB and compare
the vocal learning abilities of humans and songbirds. They look very similar not
only behaviorally but in terms of the neural and genetic mechanisms involved,
and they represent a good example of homoplasy. So this aspect of the SM system
is evolutionarily continuous and the human version is not species-specific. It is
not domain-specific, either, because the same vocal learning ability can be put to
use outside the linguistic domain. In this weak sense, it is certainly not part of
FLN. Still, human vocal learning is not exactly the same as bird vocal learning,
and it does not work quite the same for linguistic and non-linguistic purposes.
In this strong sense, the human linguistic vocal learning can be argued to belong
to FLN, too, because it has properties that cannot be found anywhere else.
Now consider Merge (recursion) as the sole candidate of FLN. The species-
specificity and domain-specificity of Merge is secured by the fact that it is the
elementary computational operation dedicated to human language only. Still,
outside the linguistic domain, Merge-like combinatorial operations can be found
in human and non-human physical/metal capacities, including locomotion, tool
using and tool making, counting, the theory of mind, mental time travel, etc.
In addition, as will be argued below, Merge is likely to have its evolutionary
Fallacies in evolutionary linguistics 147