Two talks by Jochen Zeller

Linear order affects agreement with conjoined noun phrases: experimental evidence from isiZulu“

In this talk I discuss different aspects of negation in the Bantu language isiZulu (Nguni; Guthrie code S42), which is the home language of almost  a quarter of South Africans (Census 2022). The talk begins with a brief overview of negation strategies in Bantu languages more generally, and in isiZulu specifically. I then discuss three different constructions that can be used in isiZulu to negate a transitive sentence. In the unmarked strategy, an object marker is attached to the negated verb that agrees in noun class with the object. When the object marker is omitted, the object (or the VP) is contrastively focused. In the third strategy, the initial vowel of the object (the so-called augment) is dropped, and the augmentless ([-A]) object is interpreted as an NPI. I explain how these three strategies were analysed in my previous research (Zeller 2021), and then zoom in on the properties of [-A]-nominals. Based on analyses that formulate the licensing conditions for [-A]-nominals in terms of negative concord (Carstens & Mletshe 2016; Carstens 2025) or structural case (Halpert 2015; Pietraszko 2021), I outline a proposal for a possible analysis of [-A]-nominals in isiZulu as property type arguments, which is inspired by ideas put forward in Partee & Borschev (2004) in their analysis of the Genitive of Negation in Russian.
atterns of agreement with coordinate structures whose conjuncts have different feature specifications have been shown to depend on whether the agreement controller follows or precedes the target of agreement (see Corbett 1983 and subsequent work). In this talk I report the results of four elicited production experiments conducted with speakers of the Bantu language isiZulu (Nguni; S42), in which we examined agreement with  pre- and postverbal conjoined subjects and objects. Our study finds that in isiZulu, closest conjunct agreement is a productive strategy: with conjoined NPs that appear in preverbal position, agreement with the noun class features of the last conjunct is common, while conjoined NPs  in postverbal position predominantly licence first conjunct agreement. Crucially, with respect to closest conjunct agreement, conjoined  subjects and conjoined objects in isiZulu pattern alike. We show that this result provides evidence in favor of the distributed agreement  model proposed in earlier work on conjunct agreement (Bhatt & Walkow 2013; MaruÅ¡iÄ et al. 2015), which allows for agreement to be partially established at PF, based on the linear relation between the agreement controller and the target