Syntax-Prosody in Optimality Theory
Theory and Analyses
Edited by
Jennifer Bellik [+–]
University of California, Santa Cruz
View Website
Jennifer Bellik is a postdoctoral researcher and lecturer in the Department of Linguistics at the University of California, Santa Cruz.
Junko Ito [+–]
University of California, Santa Cruz
View Website
Junko Ito is a Research Professor of Linguistics at the University of California, Santa Cruz.
Nick Kalivoda [+–]
Lund University
View Website
Nick Kalivoda is a postdoctoral researcher, Centre for Languages and Literature, Lund University.
Armin Mester [+–]
University of California Santa Cruz
View Website
Armin Mester is Research Professor and Professor Emeritus at the University of California, Santa Cruz.
Optimality Theory has become the dominant approach to studying phonology, including analyses of the mapping from syntactic structure to prosodic structure. However, when both syntactic and prosodic structures are represented as trees, it is difficult, if not impossible, to systematically generate by hand all the possible candidates, i.e., all the possible prosodic parses that must be considered in an OT investigation for any given syntactic input. Consequently, most existing syntax-prosody analyses are in this way incomplete, compromising their very validity.
This volume presents a series of studies of the syntax-prosody interface that are complete in this sense, thanks to their use of the SPOT application ( http://spot.sites.ucsc.edu ). This JavaScript application (developed by the editors) automates candidate generation and constraint evaluation, making a rigorous OT analysis of syntax-prosody possible. SPOT allows the user to test the typological predictions of the numerous proposed constraints on prosodic markedness and syntax-prosody mapping, so that researchers can make progress toward determining which formulations of the constraints should actually be part of the universal CON. A theme of the volume is comparing Match Theory (Selkirk 2011) with the older Align Theory of syntax-prosody mapping, with the finding that both are needed, at least in some languages.
Series: Advances in Optimality Theory
Table of Contents
Chapter 1
University of California, Santa Cruz
Jennifer Bellik is a postdoctoral researcher and lecturer in the Department of Linguistics at the University of California, Santa Cruz.
University of California, Santa Cruz
Junko Ito is a Research Professor of Linguistics at the University of California, Santa Cruz.
Lund University
Nick Kalivoda is a postdoctoral researcher, Centre for Languages and Literature, Lund University.
University of California Santa Cruz
Armin Mester is Research Professor and Professor Emeritus at the University of California, Santa Cruz.
Optimality Theory requires the analyst to formally define the space of possible candidates, and to consider all the candidates in that space. However, it is impossible to do this by hand in the context of research in the syntax-prosody interface, due to the complexity of the inputs and outputs, which both take the form of trees. The SPOT application automates this process, making it feasible to conduct rigorous OT analyses of issues on the syntax-prosody interface. This chapter introduces the representational assumptions of Prosodic Hierarchy Theory, as well as the key concepts of modern OT: systems consisting of fully explicit candidate sets and constraint sets, which yield factorial typologies amenable to analysis in terms of Property Theory. An overview of the book shows how this conceptual toolkit is brought to bear on major questions of syntax-prosody mapping, such as matching versus alignment, strict versus weak layering, and the nature of markedness constraints.
Part I: GEN Settings
University of California, Santa Cruz
Jennifer Bellik is a postdoctoral researcher and lecturer in the Department of Linguistics at the University of California, Santa Cruz.
SPOT Project, University of California, Santa Cruz
Edward Shingler is an undergraduate research assistant and programmer for the SPOT project, Linguistics BA 2021, University of California, Santa Cruz.
A key component of an OT system is the Gen function, which defines the candidate set for that system. In the domain of syntax-prosody mapping, the candidates are pairs. The goal of this chapter is to develop a deeper understanding of how to construct the outputs in syntax-prosody mapping candidate sets, and of how the number of candidates grows with the addition of terminals. To those ends, this chapter describes the options that the SPOT app makes available for defining Gen, and establishes through mathematical reasoning that SPOT’s Gen functions are generating the correct number of candidates for each set of parameter values. We find that the cardinalities of SPOT’s Gen functions are related to: the powers of two, the Fibonacci sequence, and the super-Catalan or Little Schröder numbers, depending on the parameterization selected.
SPOT Project, University of California, Santa Cruz
Max Tarlov is an undergraduate research assistant and programmer for the SPOT project, Linguistics BA 2020, University of California, Santa Cruz.
Much of Optimality-theoretic research is concerned with the interaction of constraints, but as the previous chapter lays out, OT analyses require a well-defined GEN in addition to a well-defined CON and EVAL. This chapter shows how different definitions of GEN (Strict vs. Weak Layering) can interact with constraints in ways that are not immediately obvious. This chapter uses two different approaches, Match Theory and Align Theory, to the analysis of Kinyambo and property analysis of the resulting typologies as a case study to this effect.
Part II: Match Theory
Lund University
Nick Kalivoda is a postdoctoral researcher, Centre for Languages and Literature, Lund University.
Irish phonological phrasing has been the subject of a number of analyses in recent years (Elfner 2012; Bennett, Elfner, McCloskey 2016, 2019). A ranking paradox noticed by Elfner 2012 remains unsolved in the context of standard OT with strict domination. We show that the problem can be solved by introducing a Match constraint sensitive only to overtly headed XPs, and by altering the definition of StrongStart to refer specifically to the left edge of the intonational phrase. The resulting analysis makes predictions for longer sentences of Irish.
PhD student, University of California, Santa Cruz
Nicholas Van Handel is a graduate research assistant for the SPOT project and a Ph.D. student, Department of Linguistics, UC Santa Cruz.
Ph.D. student, Department of Linguistics, UC Santa Cruz
Dan Brodkin is a graduate research assistant for the SPOT project and a Ph.D. student, Department of Linguistics, UC Santa Cruz.
Ph.D. student, Department of Linguistics, UC Santa Cruz
Benjamin Eischens is a graduate research assistant for the SPOT project, and a Ph.D. student, Department of Linguistics, UC Santa Cruz.
Recent work has expanded Con to include Match constraints that are sensitive to subcategories of XP and φ, e.g., Ishihara’s (2014) sp.Match(XP [+max] , φ [+max] ). Admitting subcategory-sensitive Match constraints leads to two theoretical problems: first, a proliferation of possible Match constraints, and second, the emergence of Match constraints which enforce syntax-prosody non-isomorphism, e.g., sp.Match(XP [−max] , φ [+max] ). This chapter presents the results of a SPOT investigation showing that two sets of subcategory-sensitive Match constraints drive non-isomorphism: (i) those in which only the second argument has a feature specification for its subcategory, e.g., sp.Match(XP, φ [+max] ), and (ii) those in which the first and second arguments have conflicting specifications for subcategory, e.g. sp.Match(XP [+max] , φ [−max] ). A ban on the existence of these constraints is proposed, and it is argued that this ban follows from the integration of Match constraints into the theory of Faithfulness.
PhD student, University of California, Santa Cruz
Nicholas Van Handel is a graduate research assistant for the SPOT project and a Ph.D. student, Department of Linguistics, UC Santa Cruz.
An important question in Match Theory concerns which syntactic constituents are visible to Match constraints. This chapter focuses on three proposals at the phrasal level: (i) Match(XP Lexical , φ) = Match(LexP, φ), which sees only lexical XPs; (ii) Match(XP OvertlyHeaded , φ) = Match(OhP, φ), which sees only XPs with phonologically overt heads; and (iii) Match(XP General , φ) = Match(XP, φ), which sees all XPs. Based on data from Italian, Irish, and Xitsonga, it is argued that Match(OhP, φ) usually can and sometimes must be used instead of Match(LexP, φ). This raises the question of whether the lexical/functional distinction is actually needed: although the lexical/functional distinction is a useful heuristic, because it often correlates with the silent/overt head distinction, it may not be necessary to capture the full range of phrasing data.
Part III: Align Theory
Lund University
Nick Kalivoda is a postdoctoral researcher, Centre for Languages and Literature, Lund University.
While Match Theory was intended to replace Align/Wrap Theory, Bellik et al. (to appear) have argued that both Match and Align constraints are needed to account for phonological phrasing in Japanese. Match and markedness alone cannot handle a left/right phrasing asymmetry in Japanese, while Align can. But Align cannot account for prosodic recursion of sufficient depth, and so Match is needed too. Here we analyze the resulting hybrid theory with Match, Align, and Binarity constraints. We define an OT system with these constraints, generate its factorial typology, and elucidate the relation between intensional properties of its grammars and extensional traits of its languages in terms of Property Theory (Alber & Prince 2021, to appear).
PhD student, University of California, Santa Cruz
Richard Bibbs is a graduate research assistant for the SPOT project and a Ph.D. student, Department of Linguistics, UC Santa Cruz.
Many languages allow clitic movement to certain positions, and in most cases this movement is syntactically driven. However, there are cases where this movement is prosodically motivated, most notably Chamorro (Chung 2003) and Irish (Bennett, Elfner, McCloskey 2016). Previous work has demonstrated that clitic movement may be driven by prosodic subcategorization (Chung 2003), or via constraints on prosodic well-formedness (Bennett, Elfner, McCloskey 2016). For Chamorro, clitic movement does not require prosodic subcategorization, and instead can be motivated through the interaction of syntax-prosody mapping constraints and markedness constraints on prosodic well-formedness. It will be shown that only Align constraints on syntax-prosody can motivate clitic movement, with Match constraints being insufficient.
Ph.D. student, Department of Linguistics, UC Santa Cruz
Yaqing Cao is a Ph.D. student in the Department of Linguistics, UC Santa Cruz.
PhD student, University of California, Santa Cruz
Richard Bibbs is a graduate research assistant for the SPOT project and a Ph.D. student, Department of Linguistics, UC Santa Cruz.
University of California, Santa Cruz
Jennifer Bellik is a postdoctoral researcher and lecturer in the Department of Linguistics at the University of California, Santa Cruz.
Unlike Align constraints, Match constraints cannot be satisfied by a non-recursive prosodic parse of a recursive syntactic input. As a consequence, when the candidate set is limited to non-recursive, strictly layered structures, or when a constraint on NONRECURSIVITY is undominated, Match will always be violated at least once, and can select some surprising candidates. This chapter explores the interaction of Match constraints with strictly layered candidate sets, and compares this interaction to the behavior of Align and Wrap constraints in the same context. A case study of Xiamen Chinese tone sandhi domains demonstrates that Match constraints can select superficially mismatching optima, even without the intervention of prosodic well-formedness constraints, and that neither Align nor Match constraints can capture the Xiamen Chinese pattern unless combined with prosodic subcategorization.
Part IV: Prosodic Well-Formedness Constraints
University of California, Santa Cruz
Jennifer Bellik is a postdoctoral researcher and lecturer in the Department of Linguistics at the University of California, Santa Cruz.
PhD student, University of California, Santa Cruz
Nicholas Van Handel is a graduate research assistant for the SPOT project and a Ph.D. student, Department of Linguistics, UC Santa Cruz.
Constraints on Binarity are commonly used to capture size effects: the tendency for longer strings to be parsed into more prosodic constituents. In some implementations, binarity is assessed locally by counting immediate children (= branch-counting); in others, binarity is assessed globally by counting all descendants of some category (= leaf-counting). Branch-counting binarity motivates size-driven prosodic recursion, and operates as a special case of Match(XP). In contrast, leaf-counting binarity motivates size-driven category promotion, and conflicts with Match(XP), leading to larger typology sizes. A constraint on Uniformity is shown to be able to derive size-driven mismatches as well.
University of California, Santa Cruz
Jennifer Bellik is a postdoctoral researcher and lecturer in the Department of Linguistics at the University of California, Santa Cruz.
Analyses of syntax-prosody mapping rely on notions of purely phonological well- formedness, and yet the constraints defining this well-formedness are not as clearly defined as the mapping constraints, in part because the space of possible prosodic mismatches has not been fully explored. This chapter examines several ways to define the prosodic well-formedness constraints EqualSisters (Myrberg 2013) and StrongStart (Elfner 2012, Bennett, Elfner, McCloskey 2016), and the consequences of these definitions for the predicted typology, with a focus on stringency interactions between them. Results are argued to support the use of categorical, parent-oriented definitions of both EqualSisters and StrongStart.
Tutorial
How to Use SPOT [+–]
University of California, Santa Cruz
Jennifer Bellik is a postdoctoral researcher and lecturer in the Department of Linguistics at the University of California, Santa Cruz.
Lund University
Nick Kalivoda is a postdoctoral researcher, Centre for Languages and Literature, Lund University.
This chapter contains a step-by-step guide to how to use SPOT to build violation tableaux, as well as information on how to further analyze these tableaux in OTWorkplace and other OT software.
