Volume 28, Issue 2

This work presents a heuristic mathematical approach to modeling and quantifying constraint violations. The empirical phenomenon consists of different word orders with contrastive focus in the Spanish variety of Catalonia and in Catalan. Experimental data has been collected with a gradient acceptability judgment test for auditorily presented focus constructions. The most unmarked position for contrastively focused objects in the Spanish of Catalonia is in situ or sentence-final, while in Catalan sentence-initial and in situ position is preferred. In Spanish, the subject has to be postverbal if the focus is preposed, but these constructions still remain marked. These results are explained with a model according to which each constraint has a numerical violation cost, and according to which the costs of all constraints violated by a sentence cumulate. The presumed constraint set is analyzed with an algebraic approach based on a system of linear equations with which violation costs are calculated. The methodology is heuristic in the sense that it can be used for exploring whether there could be a “missing piece”. Such a missing piece would be another constraint to be included post hoc in the set. It corresponds to another unknown added to the system of linear equations.

In this paper, I present five major Serbo-Croatian conjunctions, and argue that they differ in semantic as well as morphological complexity. I provide a compositional analysis of the morphological make-up of the arguably morphologically complex conjunctions, and establish a mapping between the two scales of their complexity. I argue that the Serbo-Croatian i is the most primitive conjunction, involving a bare addition, without any special syntactic, semantic or information-structural constraints. All other analyzed conjunctions: the disjunctive ili , the negative ni , the oppositive a and the adversative ali are more complex, they involve information-structural restrictions, and some of them also additional, polarity-related elements. For this reason, I argued that these other conjunctions are all restricted to coordinating bigger structural units: those rich enough to involve polarity- and information structure-related components, or more precisely PolPs, MoodPs or CPs. Part of the analysis is a novel view at the syntactic and semantic properties of the Serbo-Croatian clitic li , defining it as an element that pairs up an unvalued polarity feature with a focal element in its scope. One consequence of the proposed analysis is a compositional derivation of disjunctive coordination instead of taking disjunction as a cognitive and linguistic primitive.

The Prosodic Minimality Hypothesis (PMH) connects the minimal size of words to the minimal size of feet. The typological facts that bear on this hypothesis are complex, and have led some recent researchers to abandon the idea that smallest words are in fact smallest feet, instead attributing minimality to general constraints against short words. The goal of this article is to test the predictions of the PMH in the context of Optimality Theory. I argue that the PMH is empirically sound but obscured by coercion, an effect that is well-motivated and empirically supported on independent grounds. Moraicity of segments is negotiated by constraint interaction, and can interact with word minimality to produce what appear to be counterexamples to the PMH. Once this effect is taken into account, the predictions of the PMH match the typology fairly cleanly. I illustrate these arguments with detailed analyses of several languages, and with a broader look at the typological predictions of the theory.