myriad factors impacting the particulars of individual existence.) The third
factor, (more general) principles not specifi c to the faculty of language, consists
of general properties of biological, physical, and/or computational principles
of two kinds: (i) data processing and (ii) architectural/computational and
developmental constraints. By assuming (as must everyone) the existence of
the third factor in addition to the other two factors, (see also Chomsky 1965 )
the language faculty can be understood as an organ constituting part of a
biological computational system, conforming also to physical law, and as such
must be constrained by the general constraints on all such systems. The trick
(as in all empirical inquiry) is to determine the truth of the matter, i.e. for
each aspect of knowledge, how did the three factors contribute to its
internalization?
With respect to linguistic variation, Berwick and Chomsky (2011) hypothesize
that variation is, perhaps entirely, attributed to externalization:
Parameterization and diversity, then, would be mostly – possibly entirely –
restricted to externalization. That is pretty much what we seem to fi nd: a
computational system effi ciently generating expressions interpretable at the
semantic – pragmatic interface, with diversity resulting from complex and
highly varied modes of externalization, which, furthermore, are readily
susceptible to historical change.
(Berwick and Cho msky 2011)To illustrate such variation in externalization, consider the availability of overt
(main verb) V-movement famously discussed in Pollock (1989) and Chomsky
(1995), variation which was ascribed to the feature strength of Agr: If Agr is
strong (e.g. in French), verbs undergo overt movement to T. If Agr is weak
(e.g. in English), verbs stay in situ. As mentioned in the Borer-Chomsky con-
jecture, features on functional heads (e.g. strong vs.weak Agr) differentiate
individual I-languages. The deep (unifying) idea is to reduce apparent complex
diversity, to ‘small’ analytical differences. Similarly in Chomsky (2014), T is
distinguished in terms of its ability to label: Strong T can be a label (e.g. in
Italian), while weak T does not have the capacity to provide a label by itself
(e.g. in English). From this one variant property of one lexical category, cross-
linguistically variant Extended Projection Principle (EPP) and Empty Category
Principle (ECP) effects are deduced. “Italian and English appear radically dif-
ferent” but under analysis they are, by hypothesis, almost identical. These two
analyses posit crosslinguistically variant syntactic derivations executed by an
invariant universal syntax with the variant derivations triggered by language-
specifi c strength of features on functional heads in the lexicon. The nature of
the language-specifi c features is learned through experience, i.e. via exposure
(of UG) to externalized input. That is, the locus of variation is not the narrow
syntax but (a highly restricted part of) the lexicon. Consider another case, word
order variation. One example is externalization of the symmetric relation
130 Miki Obata and Samuel Epstein