On partial pro-drop patterns in Dinka

Upon greater emphasis on microparametrically-oriented research, it has become apparent that the range of crosslinguistic variation observed seems to emphatically exceed the predictions made by binary macro-parameters, and the predicted parametric clusters do not materialise consistently across all language types (see, e.g., Roberts, 2019, for relevant discussion). The Null Subject Parameter (NSP) is one such example of a parameter that escaped exclusively macro-parametric treatments, having turned out to represent a far from unified phenomenon, unlike initially assumed. Rizzi (1986)’s original binary, macro-parametric typology (predicting Consistent Null Subject Languages, Semi NSLs, and non-NSLs) quickly became too restrictive since the study of so-called partial null subject languages (PNSLs). PNSLs (in their prototypical form and simplifying grossly) usually license pro-drop for both 1st and 2nd person null subjects as well as generic 3rd person null subjects in main clauses, but disallow referential 3rd person null subjects in root contexts (e.g., Brazilian Portuguese, Ukrainian, Finnish and Hebrew).

In this talk, I discuss preliminary data from Dinka (Nilotic) that pushes the limits of variation further as regards PNSLs. Dinka is shown to present a typologically noteworthy system of partial pro-drop, which has gone largely unnoticed. In particular, drawing on a small corpus gathered from available data in the literature, I argue that Dinka likely falls within the class of inverse PNSLs (borrowing the term in Biberauer, 2018), displaying pro-drop for 3rd person in both main and embedded clauses and disallowing it for 1st/2nd person. The system takes a similar (though crucially not identical) form to languages such as Shipibo (Paoan) and earlier Germanic (see Camacho and El ́ıas-Ulloa, 2010; Walkden, 2013; Kinn, 2016), which generally also allow pro-drop only for referential third person subjects and disallowing omission of 1st/2nd person subjects in root contexts. To the best of my knowledge, though, this is the first reported language that displays pro-drop exclusively in the 3rd person in both main and embedded clauses. This system is discussed in a broader typological context, with mention to other PNSLs, such as those described in Roberts (2019) or Camacho (2013).

Although grammaticality judgements are thus far unavailable and the sample remains necessarily preliminary, the possibility that Dinka represents a system of the kind outlined above has some implications. I suggest inverse-type PNSLs of the sort attested in Dinka (and elsewhere) are hard to accommodate in several minimalist approaches to the NSP (such as Holmberg, 2010), given that they typically address PNSL systems with 1st/2nd null subjects in main clauses. After reviewing some proposed accounts of inverse PNSLs, I outline possible ways in which to address this weakness and capture Dinka’s system (e.g., by incorporating Distinct Agreement; Rosenkvist, 2010, 2018), although a precise formalisation of the patterns observed in Dinka and beyond is left for future work. I finish by arguing that such increasing range of crosslinguistic variation is most at home in so-called neo-emergentist approaches to language variation, which make minimal assumptions about the content of Universal Grammar and leave room for greater flexibility in both language variation and learnability (see, i.a., Biberauer and Roberts, 2015; Biberauer, 2019; Ramchand and Svenonius, 2014, and cf. more recent perspectives in Leivada and Murphy, 2022).