In this paper we show how Jessen & Ringen's (2002) analysis of voicing in German stops can be extended to account for the voicing of German fricatives. It is argued that while stops in German contrast for the feature [spread glottis], fricatives contrast for [voice] (and [spread glottis]). Our analysis, which involves presonorant faithfulness, is compared to an analysis with coda devoicing. We show that the two analyses make crucially different predictions, and present experimental evidence in support of the presonorant faithfulness analysis. The experimental results show considerable variation, which can be accommodated in our OT analysis.

For any phonotactic restriction on syllable onsets and codas, it can be shown that parallel restrictions are attested at edges of each higher prosodic domain. Onsets can be required at the beginnings of syllables, words or utterances, codas can be banned at the ends of any of these constituents and so on. This paper argues that these restrictions follow from constraint schemata: any markedness constraint on syllable onsets or codas (MOns or MCoda) is part of a family of constraints (MOns(Ons/PCat) or MCoda(Coda/PCat)) which imposes parallel restrictions on initial onsets or final codas of each prosodic domain. These prosodic domain-edge markedness constraints can induce epenthesis, deletion or other segmental changes at domain edges; they can also shape the prosodic structure of words.

This paper presents a previously unnoticed universal property of stress patterns in the world's languages: they are, for small neighbourhoods, neighbourhood-distinct. Neighbourhood-distinctness is a locality condition defined in automata-theoretic terms. This universal is established by examining stress patterns contained in two typological studies. Strikingly, many logically possible – but unattested – patterns do not have this property. Not only does neighbourhood-distinctness unite the attested patterns in a non-trivial way, it also naturally provides an inductive principle allowing learners to generalise from limited data. A learning algorithm is presented which generalises by failing to distinguish same-neighbourhood environments perceived in the learner's linguistic input – hence learning neighbourhood-distinct patterns – as well as almost every stress pattern in the typology. In this way, this work lends support to the idea that properties of the learner can explain certain properties of the attested typology, an idea not straightforwardly available in optimality-theoretic and Principle and Parameter frameworks.