Abstract The main aim of this study was to use the methodologies of Du Bois (1987) and Givón (1983) to investigate the relationships between various features associated with topicality (role (A, S, O), humanness, givenness, person, topic persistence (TP), referential distance (RD)) and the type and position of referential elements within the clause, in a corpus of Eastern Armenian oral narrative texts. Three main issues are explored here: factors affecting the distribution of lexical NPs , factors affecting the distribution of pronouns vs. zero anaphora, notably the ‘emphatic’ pronoun inkʿə , and factors affecting the position of objects (preverbal vs. postverbal). The results show that the most important factor favouring the avoidance of lexical NPs and the use of a pronoun or zero anaphora is low RD, with humanness, high TP, and role prominence (agent role) as probable further contributing factors. Zero anaphora and the ‘emphatic’ pronoun inkʿə are associated with low RD, while the average RD of demonstratives is intermediate between that of zero and lexical NP. Inkʿə is distinguished from all other anaphoric elements by its high average TP. Definiteness is argued to be the key factor in object position, with indefinite (especially bare) objects typically appearing preverbally, while definite objects can also appear postverbally.
Abstract Jitotoltec is the only language of the Zoquean branch that developed an overt inverse marking system. All the languages of the Mixe-Zoquean family have a hierarchical alignment, but only Mixean languages and Jitotoltec developed an inverse marking with overt inverse morphology. The language has a split inverse system in which 3:1 follows a purely hierarchical pattern, but 3:2 and 3 obviative:3 proximate follow an inverse system. In Mixean languages, the inverse marker developed from an old passive construction, a historical development that is synchronically attested in Jitotoltec. Passive morphology is a common source for the emergence of an inverse marker in languages with inverse voice/obviation or with inverse alignment. The source comes from cases in which a 3rd person O is higher in topicality with respect to a 3rd person A. Unlike a canonical passive with non-topical A, the A in the inverse continuous to be topical. Thus, an inverse construction is formally coded as a passive but pragmatically it is different from a passive. The path of development from passive to inverse is demonstrated through the comparison of the development attested in Mixean languages. On the other hand, the functional difference between a canonical passive and an inverse developed from a passive is clearly appreciated via the comparison of Topic Persistance and Referential Distance measurements obtained of the two different constructions in the study of narrative texts.
Abstract Amis (an Austronesian language of Taiwan) has a symmetrical, multi-voice system. Selection of voice and argument structure are multifactorial, with lexical, syntactic and semantic constraints. This corpus analysis of Northern Amis also shows voice and alignment to be selected by the arguments’ referential properties both within and beyond the clause. In the latter case, they react to information management and topic tracking. Discourse factors thus interact with syntax and conspire to choose the definite nominative Patient as the Preferred Syntactic Argument (PSA) argument to be tracked across clauses and sentences. The discourse function of voice selection is infrequently researched in Formosan languages; new perspectives are presented in Northern Amis.
Abstract An underlying question in linguistics is which aspects of speech might be direct reflections of cognitive factors and which might be mediated by the grammars of individual languages. One strategy for addressing such questions is to compare speech from different languages on the same topic, produced under similar circumstances. Here this approach is brought to the expression of topicality and reference in three genealogically and areally unrelated languages: Mohawk, Navajo, and Central Alaskan Yup’ik. All are head-marking, but they differ typologically in several ways, among them their pronominal distinctions, the semantic specificity of their verb stems, their alignment types, their basic constituent orders, and their clause-combining strategies. Not surprisingly, some matching patterns across the languages suggest direct cognitive effects. All show bulkier coding for less accessible referents and reduced or even zero coding for more accessible ones, and all show similar prosodic patterning in the packaging of information. But there are also differences. In some cases, existing structures are exploited to facilitate identification of referents, while in others, strategies aimed at facilitating of referent identification have apparently shaped different grammatical developments.
Abstract Differential Object Marking (DOM) is a pattern in which some grammatical objects in a language bear special marking, while others do not. Well known cases involve a range of often interacting factors: inherent properties of objects, such as animacy or person; (pro)nominality; and/or contextual properties, such as definiteness or specificity. Recent work has shed light on the role that information structure, especially topicality, plays in DOM ( Dalrymple and Nilolaeva 2011 ; Iemmolo 2010 ), but so far this line of inquiry has concentrated on topicality at the level of the proposition ( Lambrecht 1994 ). Zenzontepec Chatino (Otomanguean, Mexico) has been argued to display DOM in which objects are marked if they are (secondary) topic ( Dalrymple and Nilolaeva 2011 ). A larger and more naturalistic corpus of language use reveals that discourse topic ( Givón 1983 ), and not sentence topic, is the principal determinant of DOM in Zenzontepec Chatino. These findings broaden our view and expand our toolkit for understanding this common and intriguing, if sometimes confounding, linguistic phenomenon.
