Special characteristics of the Indo-European Dative Subject Construction in the typological landscape
A final issue that we would like to discuss is the cross-linguistic fact that if a language has non-canonically case-marked subjects, these tend to be in the dative case expressing the experiencer. Therefore, one may argue, the facts discussed here do not necessarily show that the Dative Subject Construction is inherited in the Indo-European daughter languages under discussion, as these facts might as well be explained by reference to a typological universal or a near-universal defined by a conceptual constraint that experiencer subjects be in the dative case. Given the high number of languages with dative experiencers in the world, such a conceptual constraint might seem motivated.
We have several objections to this. First of all, the "typological dative", i.e. the dative found cross-linguistically, is not cognate to the Indo-European dative, so it does not necessarily represent the same functional category or categories. Second, the question arises regarding what is meant by "dative" here, as what is referred to as "dative" in the world's languages is often a third case, i.e. neither a nominative/ subject case, nor an accusative/object case, but a third morphological marker. In other words, there is no uniform category of "typological dative", but rather a category of a third case. Let us now consider Tables 4-8 for Icelandic, Ancient Greek, Early/Classical Latin, Russian, and Lithuanian, which give the case patterns/valency patterns found in these languages. Patterns that are crossed out exist in some of the branches, but are not found in that specific language branch. Bracketed
Table 4. Case and argument structure constructions in Germanic
Table 5. Case and argument structure constructions in Latin (preliminary)
Table 6. Case and argument structure constructions in Ancient Greek (preliminary)
Table 7. Case and argument structure constructions in Lithuanian (preliminary)
patterns are patterns that we have not been able to document but may still exist. Notice that non-canonical subject marking is found for accusative, dative, and genitive, in all five language branches, even though we are only dealing with the Dative Subject Construction in this chapter. This stands in a stark contrast with,
Table 8. Case and argument structure constructions in Russian (preliminary)
Table 9. Case and argument structure constructions in Japanese
for instance, Japanese, shown in Table 9, where the case labeled "dative" is much more of a third case than in Indo-European, as accusative and genitive cannot be used as non-canonical subject markers in Japanese.
Therefore, since the Indo-European language branches have at least three different morphological cases functioning as non-canonical subject markers, there is no issue of "dative" here. That is, since non-canonical subject marking is divided across three different case categories in our five Indo-European languages, the dative case is not a specific third case, used for functions other than the ordinary subject and object functions. Hence, any claims that the use of the dative as a non-canonical subject case in the Indo-European languages is motivated by typological considerations or universal conceptual constraints are not well founded, as they abstract away from a considerably more complex situation.4 Notice also that the Indo-European languages display non-canonical case patterns that are extremely rare typologically, for instance Acc-Gen and Dat-Gen, which further strengthens our point that these argument structure constructions are inherited from a common proto-stage. This means that even though there may be a "universal" dative experiencer construction, perhaps motivated by conceptual constraints, that far from excludes that such a construction may be inherited from an earlier protostage.
The last point we want to bring up relates to the semantics of the Dative Subject Construction in our five Indo-European languages, compared to the semantics of dative experiencer constructions cross-linguistically. In particular, we want to emphasize that all five Indo-European language branches have happenstance predicates instantiating the Dative Subject Construction, although to a varying degree. This use of the dative for subjects and subject-like arguments is little motivated by any cross-linguistic considerations, although these predicates clearly denote a lack of control on the behalf of the subject referent, and in that sense they are similar to experience-based predicates. In a cross-linguistic overview of the predicates typically occurring with non-canonical subject marking, Onishi (2001) discusses happenstance predicates. However, all his examples are from Icelandic and Bengali, which are both Indo-European languages, and no examples are given from any other language family. In fact, as far as we know, it is only in research on the Indo-European languages that happenstance predicates of this type are discussed in relation to non-canonical subject marking. It therefore seems that this construction is highly specific for Indo-European, supporting our assumption that the Dative Subject Construction under investigation is inherited in the Indo-European languages.