It has been observed that a subset of dative verbs that can express causation of possession such as cwu- ‘give’, ceykongha- ‘offer’ and cikupha- ‘pay’ may be found in the double accusative frame as well as in the DAT(ive)-ACC(usative) frame in Korean. These verbs contrast with transfer of possession verbs such as kennay- ‘hand’ and phal- ‘sell’ and verbs of sending and throwing, which are found in the DAT-ACC frame only. This paper presents a meaning-based account of the limited productivity of the dative/accusative alternation in Korean dative verbs. Building on Croft et al. (2001) and Levin (2004, 2008b), I argue that the semantic classes of dative verbs form an implicational hierarchy pure caused possession > transfer of possession > caused motion, which ranks verbs in terms of the degree of the compatibility with a caused possession event type. I suggest three criteria for compatibility between verb meaning and constructional meaning and show that the analysis of verb–construction pairings proposed here, when combined with an account of variation, provides a unified explanation for verb distribution patterns observed for ditransitive constructions within and across languages and the morphosyntactic expression of recipients of dative verbs in Korean. It accounts for the limited productivity of the dative/accusative alternation in dative verbs in Korean as a consequence of choosing the cut-off point at the highest end of this hierarchy, thus explaining why only the verb class that is most compatible with the caused possession event type, i.e. pure caused possession verbs, may be used ditransitively.

This paper investigates the syntax–semantics interface within the domain of the realization of applied objects in Bantu languages, and I argue that the syntactic structure and semantic contribution of a given argument-licensing functional head (here, the applicative) do not covary. Specifically, I show that in principle, both high and low applicatives can (and should) be available with any type of applicative and not tied to a specific semantics (such as transfer of possession) or thematic role, as proposed in earlier work. Furthermore, I reject the centrality of thematic roles as a component of grammar that determines the grammatical function of applied objects, and I propose instead a typology of Bantu applied objects based on their semantic and morphological properties. This approach makes several predictions about applied objects: (i) syntactic and semantic diagnostics for high and low applicatives need not pattern together, (ii) syntactic asymmetry (such as c-command) can arise for applied objects which pattern symmetrically with other diagnostics (such as passivization), and (iii) the type of an applied object does not universally capture symmetry properties cross-linguistically. The view put forward in this paper provides a framework that can better capture this type of variation with object symmetry in Bantu languages as well as language-internal facts about applied objects; more generally, this paper sheds light on the nature of the syntax–semantic interface by showing that the meaning of a functional head is not necessarily determined by its syntactic position.

Japanese has two types of two-place motion verbs whose ‘objects’ can be marked as either accusative or oblique (accusative–oblique alternations). The accusative–goal verbs mark their objects with accusative case -o or the goal marker -ni, and the accusative–source verbs mark their objects with accusative -o or the source marker -kara. Previous studies describe systematic differences in the interpretation of the arguments of these verbs and the events they denote between the two structures. This study argues that these alternating verbs are variable behavior verbs that are linked to two distinct syntactic structures. The core evidence for this claim comes from the results of two acceptability judgment experiments with Japanese native speakers that examined: (i) selectional restrictions on the subjects of the alternating verbs and (ii) the ability of their subjects to license ‘floating’ numeral quantifiers. The results of the experiments demonstrate that the accusative–source verbs alternate between the transitive and unaccusative structures, whereas the accusative–goal verbs consistently behave like transitive verbs but assign two different structural cases to their objects. Thus, the study shows that there are multiple ways in which two-place motion verbs are mapped onto distinctive syntactic structures, whereby the core meaning of the verbs and their syntactic structures together determine their interpretation.

The likelihood with which language users insert optional words or morphemes that explicitly mark syntactic structure tends to increase in complex grammatical environments. This positive correlation between explicitness and complexity, best known as the Complexity Principle, has been observed for a multitude of case studies in both naturally occurring language and experimental settings. Researchers have sought the explanation for this Complexity Principle in three different domains: cognitive comprehension processing, the language channel, and cognitive production processing. Based on these accounts, we formulate predictions regarding the action radius of the Complexity Principle in the alternation between a direct and prepositional object of the Dutch verb zoeken ‘search’. These predictions are tested against corpus observations. Our results confirm accounts according to which optional elements indicate production difficulties, as well as those that explain the Principle as a result of restrictions on the language channel. In addition, our results indicate that the Principle is sensitive to context-determined restrictions that are the result of its underlying cause. This may present a possible caveat for alternation studies.