In a correspondence-theoretic perspective: TS is
the input, and MS the output.
TS: lz ly lx
+hr +hr -hr
-lr +lr +lr
MS: All possible structural linkers (agreement affixes, pronominal affixes, and clitics on the head, as well as morphological case on the dependent) are encoded by means of the abstract case features [+hr] and [+lr], and, possibly, by additional semantic features (such as animate, control).
Dative: [+hr,+lr]
Accusative: [+hr]
Ergative: [+lr]
Nominative: [ ]
Genitive: [+hr]/noun
By contrast, semantic case (which adds some predicate) is encoded purely by means of
morpho-semantic features.
Linking can be performed in two steps: (i) by marking
on the
head, and (ii) by case marking on the dependent.
Faithfulness constraints:
Max(+hr/+lr/f-feature): Every feature in the input has a correspondent in the output.
Ident(hr/lr/f-feature)/role: Every features values of a role are identical in the input and the output.
Surface(MS) constraints
Alignment
Markedness
Uniqueness: Each linker applies only once in a domain.
default: Every linking domain displays the default linker (nominative).
Max(lexF):
Every lexically assigned (positively-valued) feature in the input has a
correspondent in the output.
Max(+hr,+lr):
Every feature combination [+hr,+lr] in the input has a correspondent in
the output.