On the Syntax of Ditransitive Constructions

This paper deals with modelling the argument structure of constructions with two internal arguments expressing a beneficiary/recipient and a patient/theme. It offers an analysis of the dative shift which captures both the alternative grammatical function mappings and the altered semantics of the participants of the related predicates. The LMT variant used assumes that semantic participants are sets of semantic entailments of the predicate (Dowty 1991, Ackerman & Moore 2001) and that it is the syntactic representation of the predicate’s valency, rather than a hierarchy of thematic roles, that remains constant in the model (Zaenen 1993, Ackerman & Moore 2001). Specifically, instead of fixing the thematically ordered participants and allowing them to change syntactic pre-specifications (which can lead to violations of monotonicity), the proposed model keeps constant the syntactic argument positions with their fixed pre-specifications and allows the semantic participants to re-align with them. Such alternative alignments represent changes in the semantics of the predicate which are recognised when the predicate undergoes dative shift or applicative transitivisation. Since in the proposed model only those objects which are capable of becoming passive subjects are [–r] (other objects are [+o]), the model straightforwardly supports the correct prediction of the theory of object asymmetries (Baker 1988, Bresnan & Moshi 1990) that, when an argument can be a passive subject, it can also be expressed as an object marker in the active – but it does not make the incorrect prediction that the reverse will hold, too. It also concurs with Alsina’s (1996a) account of the distribution of objective properties other than passivisability; this is regulated by additional constraints which are often semantic in nature and have to be determined on a language-by-language basis. Finally, by unifying analyses of the non-applied dative and benefactive applicatives, the model provides LMT support for the special morphosyntactic status of the dative as the ‘third structural position’. 1 The argument structure of ditransitive predicates The constructions under consideration are those which are generally accepted to have two arguments in addition to the subject: a ‘recipient/beneficiary/addressee’ argument, and a ‘theme’ argument. Typical ditransitive verb meanings are ‘give’, ‘sell’, ‘bring’, and ‘tell’ (Haspelmath 2005:426), though in many languages a wide range of verbs can occur in a ditransitive valency frame with a ‘recipient’ argument. The aim of this paper is to revisit the argument structure model underlying ditransitive constructions. In particular, I offer new analyses of the dative alternation and constructions with applied recipients/beneficiaries. By focusing on the syntax of the alternating and applicative constructions, I aim to complement the recent discussion of the dative alternation undertaken by Bresnan and colleagues (Bresnan 2003; Bresnan & Nikitina 2003/2007; Bresnan, Cueni, Nikitina & Baayen 2007). Their work has stemmed from two observations: first, that Lexical Mapping Theory appears incapable of adequately accounting for the dative alternation (Bresnan 2003:19, commenting on Bresnan & Moshi 1990 and Evans 1997); and second, that the ‘classical’ form of generative syntactic theory in general does not offer appropriate apparatus to explain the gradience in the natural uses of the dative alternation 1 I wish to thank Cynthia Allen, Alex Alsina, Matthew Baerman, Dunstan Brown, Greville Corbett, Mary Dalrymple, Helge Lødrup, and Joan Maling for their very helpful questions, comments and discussion, which have led to clarifications and improvements. That is not to say that they share all the views presented here. (Bresnan 2003:71; Bresnan & Nikitina 2003/2007). In response to the second observation, Bresnan and colleagues apply a probabilistic approach to the dative alternation phenomenon, and successfully model the constraints behind the choice of the alternating variant – that is, they explain what drives the dative alternation. In the present paper, I offer a solution to the first problem by providing a theoretically satisfactory model of the dative alternation within Lexical Mapping Theory (LMT). In this way, the probabilistic approach gains an adequate lexical-syntactic representation of the modelled variants. With the revised theoretical tools, I also locate the dative alternation within the range of ditransitive constructions. Bresnan (2003:20) argues that the problem with the standard LFG account of the dative alternation (Bresnan 1978, 1982) is the assumption that the rules deriving the alternation are restricted to the information available in the lexical entries of verbs. However, the principles of function-argument correspondence proposed in LMT do exploit the semantics of lexical argument structures and allow us to tackle the polysemy of the alternating variants of ditransitive verbs. The problem lies in the fact that the most widely accepted models of LMT collapse the syntactic level of argument positions and the semantic level of thematic roles into one level of representation, which renders such models incapable of capturing any complex relationship which involves both a semantic and a syntactic alternation between related lexemes. On the widely accepted LMT analysis of the dative alternation, the arguments (identified by their thematic roles) are assumed to be the same between the two variants, therefore the same thematic roles, the beneficiary/recipient and the patient/theme, have to be pre-specified differently for the two variants to achieve the required grammatical function mappings. The following analysis is from Bresnan (2003:14-15), but see also Bresnan (2001:315), Falk (2001:113), and others: (1) a. I gave them cheques. b. gave1 〈 ag ben/rec pat/th 〉 [–o] [–r] [+o] | | |