Volume 52, Issue 2 - Differential Object Marking: Theoretical and Empirical Issues

Case marking of the object is often claimed to be driven specifically by two referential properties of the object, animacy and definiteness. Data from 744 languages, however, provide typological evidence that there is no universal preference for object case marking to be driven by these properties, but at the same time provide strong evidence that object case marking tends to be restricted in some way rather than be generalized across all objects. I argue that the independence of object case marking from these two semantic-pragmatic properties may be explained by the instability of their relationship, and that economy provides a feasible explanation for restricting case marking to only some objects.

This paper examines the interaction of DOM with information structure in Mandarin Chinese. Despite the large amount of works on this topic, much remains to be explained, in particular with respect to some alternations that do not easily fit the explanations proposed so far in terms of affectedness, animacy and definiteness. Through the analysis of text excerpts taken from the Corpus of Modern Chinese of the Center for Chinese Linguistics (CCL) of Peking University, we argue that, in addition to previously identified constraints, DOM in Mandarin Chinese performs another important function in discourse, namely that of signalling the high identifiability of the marked referent.

Many languages differentiate between different types of objects, commonly marking definite or highly animate objects. Crosslinguistically, two strategies for differentiating such objects are attested. One is differential object flagging (DOF), whereby an object is flagged by a case-marker or adposition. Another is differential object agreement (DOA), whereby the verb agrees with the object. A third strategy is to combine DOF and DOA, as happens in some North-Eastern Neo-Aramaic (NENA) dialects. This paper will focus on one such case, the Telkepe dialect, spoken in the town of Telkepe in northern Iraq. Definiteness is the main factor behind both object agreement and flagging in Telkepe: animacy does not seem to play a role. However, not all definite objects are marked as such. Objects that are generic or semantically integrated with the verb, even if activated in the previous discourse, tend not to be marked. Definite objects that are in focus also lack marking: differential object marking correlates with topic-hood, not only with the primary but also the secondary topic. These conditions for DOM are similar to Nikolaeva's (2001) findings for Ostyak. The situation in Telkepe is compared with the situation in other NENA dialects and with DOM in earlier stages of the Aramaic language. The regional parallels are also discussed, as are the possibilities of contact influence between Aramaic and Arabic dialects.

The paper deals with Differential Object Marking in Corsican. After a short introduction, it gives an overview of the main local triggering factors for marking direct objects in general (animacy, referentiality). It then presents the few main assumptions about Corsican DOM in the literature as well as findings of a new corpus study, based on written Corsican texts. Strong personal pronouns and proper names for human referents are consistently marked by the DOM marker à , but toponyms and metonymically used proper names are marked as well. Universal and negative quantifiers with a human denotation are also DOM-marked, whereas all other pronouns are not; thus animacy plays only a minor role in Corsican. The presence of determiners, quantifiers or numerals within nominals excludes the presence of à , irrespective of the nominals' denotation. Non-specific bare nominals are never DOM-marked, also irrespective of the nominals' denotation. The discussion then explains that the Corsican DOM is triggered much more by syntactic definiteness than animacy, a hypothesis strengthened by the most prominent morphosyntactic regularity at work in Corsican: nominals in combination with determiners and quantifiers cannot be marked by the DOM-marker à , even if they denote human beings. The complementary distribution of à and prenominal functional elements requires a further detailed syntactic analysis.

In this article I will outline one part of the effect of expressing semantic transitivity in Eastern Mansi. Mansi (Vogul) is a Uralic language spoken in Western Siberia. From the typological point of view, Eastern Mansi is a nonconfigurational language having case  +  agreement marking type morphosyntactic coding. The primary device for marking the direct object is verbal agreement (objective conjugation), and the function of case-marking is rather that of complementing. In this article I will discuss three main topics: 1. In what contexts does verb marking (object agreement) appear? 2. Does it appear alone or is it accompanied by a nominal object constituent? 3. When and how is the nominal object constituent marked? The theoretical basis of my study is the information structural analyses. I also make use of Differential Object Marking ( DOM ) first created by Bossong (1985) and further developed by, for example, Aissen (2003). According to my data, In Eastern Mansi, DO coding does not involve an animacy hierarchy: DO coding is primarily motivated by pragmatics. A direct object can be (1) not marked at all, (2) only verb marked, (3) both verb marked and case-marked, or (4) verb marked and marked with a possessive suffix or possessive accusative case. The variation in verb marking is based on topicality, and the role of nominal marking is to specify the object argument or to emphasize it. An object with a possessive suffix does not get any accusative marking, but with certain third person forms we use the so-called possessive accusative. Despite the earlier statements, possessive suffix does not automatically trigger the objective conjugation.

This papers offers an explanation for a seemingly inconsistent Differential Object Marking phenomenon in Komi, a Uralic language. In Komi identifiable denotations may be referred to by expressions which (i) are not specified for identifiability and (ii) in case these expression come as direct objects they may be zero marked, i.e., they take the same form as objects with non-identifiable denotations do. In practice within a specific discourse a referential object may be object marked in one instance and zero marked in the next. This kind of zero-marked objects is not predicted by prominence-based accounts on Differential Object Marking in which according to the sub-parameter of definiteness referential objects are rather expected to be accusative marked than zero marked. It is argued that in Komi object marking is triggered by identifiability marking which itself may be suppressed in contexts of givenness. Consequently, zero-marking in contrast to accusative marking can be interpreted as a givenness feature (Krifka 2007). The paper discusses this phenomenon based on observations from Komi dialectal narratives. It corroborates the distinction between identifiability as a pragmatic category and definiteness as a formal category. In showing that the dif ferent forms of Komi objects are due to their status as given (topical) vs. non-given (focal) it supports the view of information structure as a dominating parameter of Differential Object Marking.

In this paper I examine an animacy-related encoding alternation in Dutch generally observed with verbs denoting physical contact. This largely unnoticed pattern in Dutch is reminiscent of the phenomenon of Differential Object Marking (DOM) as attested in a large number of languages. I will argue that the Dutch pattern should indeed be interpreted as DOM albeit of a specific type. I claim Dutch DOM to be an instance of paradigmatic DOM used to signal a thematic difference between two groups of objects, that of animate and inanimate ones. Crucially, animate, but not inanimate, undergoers exhibit the property of sentience, a predicate entailment shown to be associated with the undergoer argument of physical contact verbs in Dutch.

This article aims at a more satisfying explanation of differential object case marking (DOM), and demonstrates that a group of mass nouns displays properties that are preserved in derivation. The central tenet of all accounts relates the Finnish type accusative-partitive DOM to the distinction between mass and count noun objects. I challenge this established view by introducing new data from Estonian: deadjectival mass nouns that unexpectedly behave like count nouns in DOM. I propose an account that has a wider coverage of data and is based on the scalar and boundedness-related properties of the base adjectives of the derived abstract nouns. Typically, the unexpected count-like behavior occurs with abstract nouns that are derived from adjectives that cannot denote open scales for various lexical-semantic and pragmatic reasons. Since the semantic properties of scales as well as the pragmatic standards determining boundedness are preserved in the course of derivation, they are cross-categorial properties. These findings are also relevant in understanding of the role of lexical aspect and aspectual composition as well as the links between morphosyntax in language and norms and standards in cognition.

At least three DOM-related constructions can be observed in Ancient Greek, but only one of them has received due consideration in the DOM bibliography (Bossong 1998). In this paper I will deal with the other two: the partitive genitive and a borderline instance of alternation in case marking of objects due to aspectual and affectedness variations in the interpretation of the predicate. I will also deal with the relation of the neuter accusative of unspecific objects with DOM, a relatively neglected construction. Central to the relation of Ancient Greek with DOM is the fact that its case marking is both universal and obligatory (or “symmetrical” in two different uses of the term). This explains why in Ancient Greek the functions differentiated by DOM are not Subject and Object, but Object and Extensions to core. Following other authors, I present here some examples of Spanish that show how the “differentiating” function (Comrie 1989) is not enough to explain DOM, even in the classical example of a DOM system, thus making clear the necessity to include constructions as the two mentioned above in a comprehensive definition of DOM.

This paper discusses differential argument realization in Abui, a Papuan language (Eastern Indonesia) with semantic alignment (in terms of Donohue and Wichmann 2008; Kratochvíl 2011). The paper examines the roles of volitionality, referential hierarchy, and specificity in differential argument realization (employing light verbs and verbal agreement) and shows that their effects are observable throughout the argument realization system and can be therefore considered the counterpart of differential case marking (DOM/DSM) reported for syntactically aligned, case-marking languages (Bossong 1983; Aissen 2003; Malchukov 2005; Kittilä 2006; de Swart 2007; and others). Although Abui differential argument realization correlates with the referential hierarchy (Bickel 2008), verbal subclasses (in terms of Tsunoda's 1985: 388 affectedness hierarchy) determine whether differential argument realization appears or not (cf. von Heusinger and Kaiser 2011). The paper also considers the diachronic origins of differential argument realization in Abui. Most constructions are argued to originate from two-clause constructions, such as topic/focus constructions gradually merging into singleclause structures. Information-structure related morphemes (light verbs in Abui) serialize with lexical verbs. The process starts with 1st and 2nd person participants (interlocutors) and may extend to human or human-like third persons. Grammaticalization ends in the fusion of the light verb and person prefix, creating new verbal agreement paradigms, as manifested in several central Alor-Pantar languages (Kratochvíl et al. 2011).