Double Object Constructions and Scrambling
The term ‘Double Object Constructions’ is ambivalent. Sometimes it refers to all phrases projected by a verb that has two object slots (typically filled with arguments having the semantic status of Patient and Addressee/Recipient). There is, however, a more restrictive use of the term. In many studies in generative syntax double object constructions (DOC) are opposed to the so called preposional ditransitive constructions (PDC). English John gave Mary the card is a DOC, while English John gave the card to Mary is a PDC. Some authors argue that even in languages, where one deals with permutable sequences of two prepositionless objects (Verb … [Obj1 + Obj2] ~ Verb …[Obj2 + Obj1]), only one constituent order signals a DOC construction, while the alternating order signals a PDC construction. Such an analysis implies that verbs, that open two object slots, allow two different syntactic configurations linked with different word order: in the DOC regime, a two-place verb (or a three-place verb, if one counts the external argument) has two object arguments of the same rank, while in the PDC regime one of the arguments is placed higher (it roughly corresponds to the notion of ‘direct object’ in traditional grammar), while the other argument stands lower.
Since Pylkkaenen (2002) it has been fashionable to claim that DOC constructions have applicative characteristics, while their pairings, the PDC, lack them irrespective of the fact, whether they are really marked with PPs in a given language or not. A theory basing on this assumption is attractive, since both the change of worder and the shift from PDC to DOC may be analyzed in terms of argument raising. A question arises, however, what’s the actual role of word order is and whether all world’s languages are alike with respect to DOC/PDC theory. Some languages (not so few indeed) are unproblematic on the reason that they completely lack ditransitive verbs. Other languages allow two prepositionless objects only with one order and ban the inversion: on this reason, they are unproblematic, too. Languages with object scrambling are problematic. One has however succeded in finding subtle difference between the alledged DOCs and PDCs in a number of Slavic and Balcanic languages, such as Modern Greek or Bulgarian: some authors, as Anagnastopoulou (2003) or Slavkov (2007) claim that DOCs and PDCs may be identified regarding such tests as possibility of clitic doubling and scopal changes. If these descriptions are correct, they confirm the speakers’ intution that different linear orders with ditransitive verbs have different communicative (or : ‘semantic’) load and imply that one of these orders is perceived as basic, and another one as derived. This result is relevant for non-generative studies, too, say, for Greenbergian typology. Therefore nothing may prevent us from assigning the labels DOC and PDC to the corresponding word orders.
Some languages with an unbounded scrambling as, e.g. Modern Russian and many present-day and some old languages with a similar combinatorics (take, for instance Old Norse), do not give us a hint beforehand, which linear order is basic and which one is derived. In a situation where both competing orders are equally possible, communicatively neutral and have nearly the same frequence, even the most trained speakers probably lack an intution, which linear order to choose as the basic option: such an knowledge gets accessible only by syntactic analysis. One can certainly ascribe ‘covert applicative morphology’ to Russian on the premise that is a feature of UG; one can also claim that an accusative Patient-like object has a priority, since accusative stands in Case Hierarchy higher than dative. But such aprioristic claims do not save us from finding independent evidence for the existence of a priviledged argument and a priviledged word order.
See a syntactic procedure for getting the basic word order in Russian ditransitive clauses in my 2007 paper published in: Текст. Структура и семантика // Доклады XI международной конференции. Moscow: 2007, 138-151. The final result – Dative-Recipient arguments normally precede Accusative-Patient arguments – is traditional, but the implications of this analysis and the syntactic tests proposed there may seem paradoxical. If you have heard that in Russian and other scrambling languages with phrasal prosodies the main phrasal accent (the focus / rhem e accent) ceteris paribus falls on the last constituent or word form, forget this myth or modify your conception!