Reconstructing constructional semantics
The dative subject construction in Old Norse-Icelandic, Latin, Ancient Greek, Old Russian and Old Lithuanian*
Johanna BarcSdal, Thomas Smitherman, ValgercSur Bjarnad6ttir, Serena Danesi, Gard B. Jenset and Barbara McGillivray*
Ghent University / University of Bergen / Stockholm University / Oxford University / Oxford University Press
As the historical linguistic community is well aware, reconstructing semantics is a notoriously difficult undertaking. Such reconstruction has so far mostly been carried out on lexical items, like words and morphemes, and has not been conducted for larger and more complex linguistic units, which intuitively seems to be a more intricate task, especially given the lack of methodological criteria and guidelines within the field. This follows directly from the fact that most current theoretical frameworks are not construction-based, that is, they do not assume that constructions are form-meaning correspondences. In order to meet this challenge, we present an attempt at reconstructing constructional semantics, and more precisely the semantics of the Dative Subject Construction for an earlier stage of Indo-European. For this purpose we employ lexical semantic verb classes in combination with the semantic map model (Barðdal 2007, Barðdal , Kristoffersen & Sveen 2011), showing how incredibly stable semantic fields may remain across long time spans, and how reconstructing such semantic fields may be accomplished.
Constructional meaning is considerably more abstract than lexical meaning, and hence more difficult to study. Using introspection to study constructional meaning may be an adequate procedure within synchronic linguistics, where the aim is to study one's own variety of language. However, introspection is not a useful analytical tool when studying abstract constructional meaning in dead languages or earlier language stages. In the present chapter we lay out how constructional meaning may be studied historically, within the framework of Cognitive Construction Grammar, based on the lexical semantics of the predicates instantiating an argument structure construction, in combination with the Semantic Map Model (Croft 2001, Haspelmath 2003, Cysouw, Haspelmath & Malchukov 2010).
Traditionally the meaning of case markers has been studied irrespective of the constructions they instantiate, and an abstract meaning has been posited for a given case, for instance the dative case, on the basis of its occurrence in a variety of different constructions (cf. Hjelmslev 1935, Jakobson 1936). This view of case is not only common for traditional grammar but it is also the prevalent view within Cognitive Grammar (Dabrowska 1997, Janda 1993). On a constructional approach, however, where constructions, i.e. form-meaning pairings, are taken to be the basic units of language, case markers constitute a part of the form side of a construction (cf. Bar8dal 2001a: 33-39, 2009, Fried 2005), and should not be studied outside of their respective constructions. Instead, they are viewed as an integrative part of the form side and should be studied as a part of that whole. On a constructional account, therefore, the view that a form may have an all-embracing meaning, generalizing over all its instances, appears to be detached from linguistic reality and the psychological reality of the speaker.
Since the assumption that constructions are form-meaning pairings is specific to the constructional framework, and not shared by other current linguistic frameworks, this also explains why no attempts to reconstruct the meaning of abstract constructions are found in the literature. The present article, therefore, is an attempt to outline a methodology on how to reconstruct the semantics of such abstract constructions, illustrating how the framework of Cognitive Construction Grammar is needed to carry out this enterprise. Hence, we will compare the semantic fields occupied by dative-subject predicates in Old Norse-Icelandic, Ancient Greek, Latin, Old Russian, and Old Lithuanian. Systematic investigation of the semantic scope of the Dative Subject Construction in each of these Indo-European languages will be carried out as a part of a larger Indo-European comparison, with the aim of casting light on the relation between verbal semantics and non-canonical case marking. A secondary goal is to study the development of the Dative Subject Construction in Indo-European and whether the construction may be reconstructed for an earlier proto-stage. We will concentrate here on the semantics of the construction irrespective of the different forms found for different subconstructions of the Dative Subject Construction in the Indo-European languages under investigation (see Section 2).
It is the general view in the literature on non-canonical subjects in the Indo-European languages (Zaenen, Maling & Thráinsson 1985, SigurSsson 1989, Falk 1997, Haspelmath 2001, inter alia) that such non-canonical subject marking is associated with experiencer and benefactive predicates. As shown by Barðdal (2004) in her comparative study of Modern Icelandic, Modern Faroese, and Modern High German, this is an oversimplification, and several dative-subject predicates in Modern Icelandic are neither experiencer nor benefactive predicates, but denote some sort of non-volitional, often accidental, events, referred to here as happenstance events. The same is true for Modern Faroese and Modern German, although there are fewer happenstance predicates in those languages than in Modern Icelandic. It remains to be documented whether the situation is similar in, for instance, Old Norse-Icelandic, the predecessor of Modern Icelandic, and in the other ancient and archaic Indo-European languages, although a preliminary comparison between Old Norse-Icelandic and Modern Icelandic reveals that the frequency of happenstance predicates in the Dative Subject Construction has gone down drastically in comparable texts (Barðdal 2011).
It has often been assumed that while the Dative Subject Construction is robust in Modern Icelandic, the same is not true for the older stages of Germanic or the Indo-European languages. Hock (1990) documents only a few potential predicates for Sanskrit, as does Luraghi (2010) for Hittite. However, our results reveal a large number of common sememes across the Indo-European language branches investigated here, as well as a major overlap in the semantic fields occupied by the construction across Germanic, Italic, Greek, Slavic, and Baltic, which may suggest an Indo-European inheritance. For this purpose, we will develop a method to reconstruct a common semantic space, on the basis of existing lexical semantic verb classes across related languages, a method which can generally be used to reconstruct constructional semantics for earlier language stages and dead languages.
In Section 2 we give a definition of our notion of the Dative Subject Construction, illustrated with examples of the various subconstructions of the construction. In Section 3 we review and discuss some of the problems with reconstructing semantics, and suggest a method for reconstructing constructional semantics on the basis of narrowly-defined lexical semantic verb classes and the Semantic Map Model. In Section 4 we compare the predicates instantiating the Dative Subject Construction in all five Indo-European language branches under investigation and suggest a classification of the relevant predicates into narrowly-circumscribed semantic classes. We also discuss the different kinds of semantic overlaps expected, depending on whether a category is inherited or not, whether the inheritance is early or recent, and whether the construction has been productive or not in the history of these languages. We then, in Section 5, attempt a comparison of the semantic space of the Dative Subject Construction across the five Indo-European branches under investigation, and on the basis of that put forward a reconstruction of the semantics of the Dative Subject Construction for an earlier Indo-European proto-stage. In Section 6 we discuss how the Dative Subject Construction in the Indo-European languages under discussion has its own specific Indo-European characteristics, which makes it different from an alleged universal "dative experiencer" construction, again corroborating our claim that the Dative Subject Construction in the Indo-European daughters is inherited from an earlier proto-stage. In Section 7 we summarize the content and conclusions of this chapter.
-  We thank the audience at the SLE workshop on "Theory and Data in Cognitive Linguistics" in Vilnius in September 2010 for their comments and discussions, as well as the audiences in Reykjavik (2011), Copenhagen (2011), Wassenaar (2012), Bergen (2012), Uppsala (2012), and Reykjavik/Eyjafjallajokull (2012). We are also grateful to Thorhallur Eythorsson, Eystein Dahl, Tonya Kim Dewey, Stephen Mark Carey, Ilja Serzant, and Valentina Tsepeleva for further comments and discussions, Misumi Sadler for help with the Japanese data on which Table 8 is based, and finally the reviewers and editors of this book for their contribution to the increased quality of the final product. Needless to say, all shortcomings are entirely ours.
-  One exception to this is Jónsson (1997-98, 2003).