The paper discusses possible scenarios on the development of the particle wo that occurs regularly in relative clauses in Upper German dialects. We claim that wo originates from the equative particle so – despite its homophony with the w-version of the locative adverb. So is the equative particle whose usage as a relative particle was fairly widespread in Early New High German. We will demonstrate that the semantics of the equative particle is well suited for its function in relative clauses. The scenario sets Upper German dialects in a row together with (many) Scandinavian languages/variants, in which the relative particle is som – again an equative particle that is historically directly related to so. It will be shown that a unified analysis of the syntax of relative clauses in the two variants of Germanic is plausible in that both variants show the same relevant properties.

There are not many uncontroversial findings regarding secondary stress in German since phonetic evidence is hard to find. In this contribution it is argued that main stress distribution in artificial words constitutes indirect proof of the existence of rhythmic secondary stress requirements in German. An analysis is proposed in terms of rhythmic and prosodic constraints which are then applied to past participle formation in German.

Der Erwerb einer Zweitsprache als Erwachsener unterscheidet sich qualitativ vom Erwerb der Erstsprache. Studien zum frühen Zweitspracherwerb bei Kindern zeigen, dass sich die Spracherwerbsfähigkeiten bereits im Alter von 3 bis 4 Jahren verändern. Die vorliegende Longitudinalstudie untersucht über einen Zeitraum von zwei Jahren den Erwerb der deutschen Partizipflexion bei vier türkisch-deutsch sukzessiv bilingualen Kindern mit einem Alter bei Erwerbsbeginn von ca. 3 Jahren. Die Ergebnisse zeigen, dass sukzessiv bilinguale Kinder die Partizipflexion wie monolinguale Kinder im Hinblick auf Fehlerarten und -umfang erwerben. Eine Ausnahme bildet der Erwerb des Präfixes ge- : Die sukzessiv bilingualen Kinder lassen das Präfix nur selten aus, was an ihrer fortgeschrittenen prosodisch-phonologischen Entwicklung liegen könnte. Die für den monolingualen Erwerb asymmetrische Fehlerverteilung bei regulären und irregulären Partizipien tritt auch hier auf: Während das reguläre Suffix -t häufig übergeneralisiert wird (geschlaft statt geschlafen), kommt der umgekehrte Fall nur selten vor. Inwieweit single oder dual mechanism-Ansätze diese Asymmetrie besser erklären können, wird seit langer Zeit kontrovers diskutiert. Die signifikant höhere Übergeneralisierungsrate des Suffixes -t bei irregulären Partizipien kann nicht auf eine höhere Frequenz von regulären Partizipien zurückgeführt werden. Dies widerspricht dem single mechanism-Ansatz und unterstützt den dual mechanism-Ansatz.

The German writing system is typically described as using two kinds of elementary units: graphemes, which are abstract entities, and graphs (or glyphs), which are their material expressions. Although this dichotomous model may be sufficient for orthographic tasks, it proves inadequate for mapping the writing system as a whole, for which it is neither differentiated enough nor consistent. This paper aims to enhance the customary model by demonstrating that the grapheme has to fulfill three incompatible systematic functions at once and, as a consequence, by establishing distinct levels within the model. In terms of graphemics, it is pointed out that although the grapheme can only be defined either as a phoneme correspondent or as a minimal semantically distinguishing unit, both contradicting definitions are often jumbled (for instance by using statistical criteria to determine minimal units). In terms of graphetics, it is argued that prototypical character shapes are systematically relevant entities as some graphemes show more than one prototypical shape. Furthermore, the choice between shapes can to some extent be influenced by morphological, lexical or other linguistic parameters, such as text length or region. As a result, the structural model of the writing system is extended to four hierarchic levels of elementary units, tentatively called "phoneme correspondents", "graphemes", "basic shapes", and "graphs". Even though the model is developed with respect to German, quintessential parts of the argumentation can be applied to other languages which use the Roman alphabet.