9 It can be argued that there are appropriate discourse features before Transfer,
but it is more adequate as the model of thought to exclude any interference
from the Communicative Interface.
10 In van Trijp (2012), the rise of case is regarded as one of the examples of self-
organization, but he does not mention the loss of case infl ection. In this chapter,
the rise and loss of case infl ection are explained in terms of the balance between
computational effi ciency and communicative effi ciency at LoC.
11 Freidin and Sprouse (1991) argue about subjecthood on the basis of the empiri-
cal evidence such as anaphoric bindings, inversion in interrogative sentences and
subject raising.
12 See Jespersen (1927: 11.2) for the traditional assumption.
13 Allen also points out some examples of conjunction reduction in OE, which
leads us to the possibility that PredP mentioned later had already emerged in
Late OE.
14 Other proposals about a thematic hierarchy include the following:
Agent > Theme > Goal > Obliques (manner, location, time,.. .) ( Larson (1988))
Agent > Goal > Theme ( Jackendoff (1972), Grimshaw (1990))
Agent > Theme > Location ( Baker (1997))15 Sigurðsson (2012:193) states, “Event licensing entails θ-licensing. While thus
being syntactically licensed, θ-roles are interpreted or read off at the C-I interface,
on the basis of structural and lexical information transferred from syntax.”
16 As the relation between semantic features and morphological cases is not universal
but idiosyncratic, (19) should be regarded as an instance of Accusative
languages.
17 Assuming the hierarchy of Nom > Acc, we can give a theoretical explanation to
Burzio’s generalization (B urzio 1986: 178–186) that case is assigned to the
object iff a θ-role is assigned to the subject. As for the unaccusativity in (i), theme
is made visible by Nom in (ia) whereas theme must be made visible by Acc in
(ib) because Nom has a priority to make the agent visible.
(i) a. dass der Fehler gefunden wurde/war
that the mistake-NOM found was/was
‘that the mistake was found’
b. dass er den Fehler gefunden hat
that he-NOM the mistake-ACC found has
‘that he has found the mistake’ (adapted from Haider (2010: 276))
Moreover, as for a nominative object in Icelandic in (ii), as there is no agent in
the sentence, the theme is made visible by Nom.
(ii) Jóni líkuðu þessir sokkar
Jon.DAT like.PL these socks.NOM
‘Jon likes these socks.’
18 See Hosaka (2013) for the derivation of the double object construction.
19 The concept of exaptation can be attributed to evolutional biology by Gould
and Vrba (1982). Owing to Lass (1990), the explanation by exaptation prevailed
even in the fi eld of historical linguistics. The proposal here is based on the
assumption that linguistic structures as well as lexical and morphological phe-
nomena are vulnerable to exaptation.
20 In Tanaka (2010) the rise of PredP in English infi nitival constructions is discussed
and Bowers (2002) argues about the existence of PredP in Icelandic.
21 As explained in Li and Thompson (1976), from the typological point of view,
languages in the world are divided into Topic Prominent Languages such as
210 Michio Hosaka