Advances in Biolinguistics - The Human Language Faculty and Its Biological Basis

(Ron) #1

(8) Merge(α, β) counts as Internal Merge (IM) if one of α, β is a term of the
other. Otherwise, Merge(α, β) counts as External Merge (EM). (Chomsky
2004 et seq.)


For some reason, endocentric {H, XP} structures like (7) are typically generated
by EM.
Then, how about SOs generated by IM? Take a representative case of IM in
(9), where the subject nP in (9a) A-moves into the “Spec-T” position, leaving
a copy of nP behind, as in (9b).^2 In this structure, the subject nP and the head
T of YP undergo φ-feature-agreement, valuing the unvalued φ-features of T (in
what follows, we will adopt the familiar notation where [uF] stands for a formal
feature F whose value is unspecifi ed, and [vF] stands for a valued, and hence
inherently interpretable, F).


(9) a.
T[uφ] vP


......
n[vφ] NP


b.

n[v ] NP T[v ] vP

... tnP...


φφ

Notice that there is no obvious sense in which such an SO is “endocentric,”
so long as we stay away from projection-based stipulations: The output of IM
is of the form {XP, YP}, where both of the constituents are phrasal, and there
is no single LI that structurally stands out as the head. Further, the structure
feeds various semantic effects at SEM, such as the subject-predicate relation and
“aboutness,” but none of these interpretations is obviously “T-like” or “n-like.”
More generally, the semantics of internally merged SOs, traditionally referred
to as “s-structure interpretations,” involves certain “discourse-related” properties
that cannot be readily attributed to any single LI or its lexical features, hence
not endocentric. The same point can be made for cases of A’-movement like
(10), where a wh-element in (10a) moves into Spec-C, accompanying some sort
of Q(uestion)-feature-agreement (see Cable 2010, Narita 2014), as in (10b).
In (10) too, we cannot attribute the whole semantics of the wh-question to any
single LI: its interrogative force can be said to be C’s, while the operator-variable
relations and the WH-quantifi cation are primarily due to which. Again, the
semantics is not endocentric and highly discourse-related.


12 Hiroki Narita and Naoki Fukui

Free download pdf