Towards a typology of middle voice systems

The middle voice is a notoriously controversial typological notion. Building on previous work (e.g. Kemmer, Suzanne. 1993. The middle voice. Amsterdam & Philadelphia: John Benjamins), in this paper I propose a new working definition of middle markers as inherently polyfunctional constructions which are partly associated with valency change in opposition to bivalent (or more) verbs and partly lexically obligatory with monovalent verbs. Based on this definition, the paper undertakes a systematic survey of 149 middle voice constructions in a sample of 129 middle-marking languages. Evidence from the sample shows that middle voice systems display a much richer variation in forms and functions than is reported in the literature. This richer empirical evidence challenges some of the mainstream views on middle marking, especially its purported connection with reflexivity and grooming-type events, and calls for an overall rethinking of the typology of the middle voice.

Predicting voice choice in symmetrical voice languages

Western Austronesian symmetrical voice languages exhibit at least two basic transitive constructions. This paper investigates what factors influence speakers’ choice of one voice over another in natural spoken discourse. It provides a thorough assessment of all factors that have been proposed to be relevant for voice choice in the literature on symmetrical voice systems. Using the Indonesian language Totoli as a case study, we show that unlike in languages with asymmetrical voice alternations, argument-related properties such as topicality, activation state, animacy, etc. do not play a major role in voice choice in symmetrical voice languages. We argue that for symmetrical voice alternations other factor groups are relevant than for asymmetrical voice alternations and that the clear structural differences between the two alternation types are mirrored in functional differences.

Alignment shift as functional markedness reversal

In this paper, we propose treating alignment shift as a process of functional markedness reversal in the domain of semantically transitive constructions. We illustrate how this approach allows us to capture similarities between the alignment shifts in Eskimo-Aleut and Western Austronesian languages, despite morphosyntactic differences in their voice systems. Using three diagnostics of functional markedness (semantic transitivity, topic continuity of P, and discourse frequency), we compare antipassive and ergative constructions in Eskimo-Aleut varieties and actor voice (av) and undergoer voice (uv) constructions in Western Austronesian varieties. We argue that ergative alignment is equivalent to a functionally unmarked P-prominent construction (e.g., ergative, uv), whilst accusative alignment is equivalent to a functionally unmarked A-prominent construction (e.g., antipassive, av). On this basis, we claim that both language groups are undergoing a parallel shift from ergative to accusative, since A-prominent constructions are functionally marked in more conservative varieties, but lose their functionally marked character and begin to function as unmarked transitive constructions in more innovative varieties.

“Could you please repeat that?” Speech design best practices for minimizing errors

Until recently, designing the interactions between systems and users has relied primarily on visual communication. However, with the dramatic increase in the use of virtual assistants, voice interaction design is now a sought after skill for human factors and user experience practitioners. In this practice-oriented paper, we aim to identify the best practices for dialog designers striving to minimize errors in communication between the user and the voice-enabled system. Furthermore, we highlight the technology behind voice systems as well as the importance of minimizing errors when our voice-controlled devices become a part of highly sensitive interactions: specifically, when accessing critical, private information. Summary: This is a practice-oriented paper highlighting six best practices for avoiding errors when designing voice-enabled systems. Errors in voice systems have been found to require a great deal of user effort to resolve, as well as strongly influence user satisfaction with their voice system. We discuss the technical foundation for voice user interfaces (VUIs) as well as the evolving nature of these systems. The purpose of this paper is to aid and inform upcoming voice design practitioners in the best method for dialog designing.

Ergativity and Austronesian-Type Voice Systems

In an Austronesian-type voice system, one argument is designated the "pivot" in each clause, with verbal morphology indicating the choice of pivot and corresponding changes in case marking and extraction possibilities. It has been suggested that ergativity plays a crucial role in these systems. This paper argues that voice and case reflect separate but interacting systems, which can be dissociated from each other, based on the behavior of voice in Dinka (Nilotic; South Sudan) and Balinese (Austronesian; Indonesia). These languages exhibit familiar voice morphology, but are shown to not involve any ergativity. Instead, we propose that what unifies "voice" system behavior is a lack of structural licensing for subjects in Non-Subject Voices. Different voice systems solve this problem in different ways: Balinese licenses the subject under adjacency with the verb, whereas Dinka allows for a last resort genitive case like in many Austronesian languages of the Philippines and of Taiwan.