Chapter 1. Investigating information structure in lesser-known and endangered languages

Abstract

This paper proposes an analysis of Austronesian cleft structure which is parallel to basic word order derivation in VOS Austronesian languages. In basic VOS word order, the subject or absolutive DP moves to a low topic position in the left periphery, and the remnant TP moves to a higher focus position. In a cleft, it is the presupposed relative clause which functions as the absolutive and moves to the topic position, while the focused constituent is pied-piped within the remnant TP to the focus position. Accordingly, this analysis is in agreement with the tradition in Austronesian syntax of analyzing the clause-initial focused constituent as either forming the matrix predicate or being contained within that predicate. This paper further agrees with the mainstream approach in proposing that this derivation is parallel to the derivation of VOS basic word order. The traditional analysis of Austronesian clefts marks a departure, however, from cross-linguistic approaches to cleft structures in which the focused constituent is analyzed as the matrix subject. This discrepancy is accounted for in the present paper by showing that the movements involved in the derivation target the CP layer, specifically the focus and topic positions in the left periphery where the moving constituents receive their respective interpretations. Consequently, neither the focus nor the headless relative clause resides in a subject A-position. Rather, both are located in A'-positions. IntroductionThis paper proposes an analysis of the structure of clefts in Austronesian languages. Cleft constructions have been addressed rather frequently in the literature, particularly because of the fact that wh-questions are formed on clefts in a great number of Austronesian languages (Georgopoulos 1991;Paul 2001;Pearson 2001;Massam 2003;Aldridge 2002Aldridge , 2004Potsdam 2006Potsdam , 2007Potsdam , 2009 among others). In the following Tagalog examples, (1a) shows basic VSO order in a declarative clause. (1b) shows a wh-question in which the direct object is the whphrase. This constituent appears in clause-initial position, but it does not reach this position through movement to [Spec, CP]. Rather, the wh-word acts as the matrix predicate in a cleft construction. The remainder of the clause is packaged as a headless relative clause preceded by the absolutive case-marker. This structure is parallel to the specificational pseudocleft shown in (1c).* I am indebted to Thess Savella and Raph Mercado for invaluable native speaker input on many of the Tagalog sentences. I also thank the participants and organizers of the AFLA 17 meeting for comments on the oral presentation of this paper.

Lexical anaphora : a corpus-based typological study of referential choice

“…Elith et al 2008;Strobl et al 2009;Ridgeway 2020). 7 The relationship between the numerous predictors in these kinds of models is highly complex: Most exert only small influence on the results, but all contribute in some way, so that removal of any one factor leads to an appreciable reduction in predictive accuracy (Loukachevitch et al 2011).…”

“…Crucially, only those clauses that have arguments coded identically to an A and a P argument are deemed transitive, though neither needs be overtly realized. 7 Andrews (2007: 139) distinguishes the syntactic functions S, A, and P from grammatical relations such as subject and object. The relation between syntactic functions and grammatical roles remains a topic of controversy in the literature, but a substantial body of research suggests that they do represent a valid level of syntactic description, and, more importantly, provide a framework within which significant cross-linguistic generalizations on the possible shapes of grammars can be formulated (Comrie 1989;Farrell 2005;Andrews 2007;Haspelmath 2011; among many others).…”

“…the rate at which non-lexical forms are incorrectly predicted as lexical). 7 Here, the dashed diagonal line indicates the curve of a so-called no-skill classifier, which always predicts the majority class. The optimal classification threshold is again the point on the ROC curve that is closest to the top-right corner of the graph, which indicates a model with perfect skill, that is one that always predicts outcomes correctly.…”

“…forms a chain of length l = 1), it is referred to as an "unchained mention", or simply a "one-off". 7 Note that this definition of clause chaining differs from the more nuanced ones in Givón (2017) and related work on discourse coherence, in that it takes an extremely simplified structural view, aiming to be as readily operationalizable and quantifiable as possible, given adequately annotated data. See also Torres Cacoullos & Travis (2019: 672-673) for further semantic refinements in defining same-subject chain contexts.…”