However, language in use is full of imperfections, which are the properties of
LoC required by the Communicative Interface, and SEM2 is solely related to
the interpretation of the output language. The semantic ambiguity occurs there,
and the duality of semantics that Chomsky refers to is attributed to the differ-
ence between SEM1 and SEM2.^8 (9a) and (9b), for example, seem to be
synonymous, but (9a) is an appropriate answer to What did John do? whereas
(9b) is to What did John do with my book?
(9) a. John gave your book to Mary.
b. Your book, John gave _____ to Mary.
(Rizzi (2011: 2 ))
Therefore, there is attested semantic difference between (9a) and (9b). If
such topicalization occurs after Transfer, it is impossible to yield the correct
output to the CI interface, which requires us to presume different semantic
interfaces at CI and SM.^9
Then, what exists in the process of LoC? There are assumed to be morpho-
phonological realizations, movement and functional projections, which are required
by the Communicative Interface and, susceptible to historical change, yield lan-
guage variation. In addition, the mechanism to produce such diversity is related
to not only language itself but also to the universal factors to produce principles
and constraints in common, which can be regarded as self-organization.^10
4 Quirky subject revisited
The problem of case in Icelandic is a daunting challenge to the generativists.
Notice that not a nominative noun but a dative noun occupies the subject
position of each sentence, as exemplifi ed in (10).
(10) a. Jóni líkuðu þessir sokkar
Jon.DAT like.PL these socks.NOM
‘Jon likes these socks.’
b. Þeim var hjálpað
them.DAT was.SG helped
‘They were helped.’
c. Um veturinn voru konunginum gefnar ambáttir
In the.winter were.PL the.king.DAT given slaves.NOM
‘In the winter, the king was given (female) slaves.’
(Bobaljik 2008: 298)
In addition, it is proved through several syntactic tests that these are not
topic and focus positions but subject positions.^11 Then it is obvious that the
movement of the dative to the subject position in (10) is not driven by case.
Therefore, in the present framework of the Minimalist approach it is assumed
that the dative moves there in order to check the EPP feature.
202 Michio Hosaka