Volume 48, Issue 4

The article aims to demonstrate the gradualness of grammaticalization of (what are traditionally regarded as) desinential case exponents in selected Finno-Ugric languages: Hungarian, Finnish and Estonian. Taking into consideration the specificity of the Finno-Ugric desinential case markers, the author develops a list of parameters which make it possible to fix the relative grade of grammaticalization of the analyzed linguistic entities. The computations show the heterogeneous and simultaneously dynamic nature of the desinential case category in each investigated language; one can trace in them the presence of more and less grammaticalized case endings which build up a kind of semi-continuum, i.e. there are no large gaps between different stages of grammaticalization. The diagram depicting the number of desinential cases occupying respective positions on the grammaticalization scale resembles the Gaussian distribution; i.e. the extrema are occupied by a relatively small number of cases, the majority of them being grouped in the middle of the scale.

In this work I will provide direct evidence against the identity of relative pronouns and complementizers/subordinators – proposed in recent works (Kayne 2010b) within the generative paradigm – with the aid of (diachronic and synchronic) data from Akkadian (an extinct Semitic language of Mesopotamia), Sogdian (an extinct Middle Iranian language), Germanic languages, Eastern and Western Iranian languages and Creoles. I will also show that the mismatch of relative (demonstrative) pronoun and complementizers does not weaken a proposal of a unified syntactic structure underlying the two clause-linkage phenomena of complementation and relativization. I will try to demonstrate that subordinate clauses (relative and complement clauses) are headed by light nouns/pro-nouns acting as "bridges of features" between matrix and dependent clauses. In particu-lar the simultaneous presence of demonstratives / relative pronouns and complementizers signalling clauses' edges in many languages is the primary evidence that (light) nom-inal elements are possibly required to trigger phenomena of clause linkage.

Based on real-text corpora with syntactic annotation, this study quantitatively addressed the following two questions: whether quantitative methods and indexes can point to the diachronic syntactic drifts characterizing the evolution from Latin to Romance languages and whether these methods and indexes can provide evidence to evince the shared syntactic features among Romance languages and define them as a distinctive language subgroup. Our study shows that the distributions of dependency directions are suggestive of positive answers to the above two questions. In addition, the dependency syntactic networks extracted from the dependency treebanks reflect the degree of inflectional variation of a language, and the clustering analysis shows that these parameters, in spite of some imperfections, can also help differentiate Romance languages from Latin diachronically and from other languages synchronically.

The present study examines how Swiss German learners cope with the contrast between voiced and unvoiced obstruents in L2 French. The feature [±voice]) is not exploited in Swiss German dialects, where pairs of obstruents sharing the same place and manner of articulation are basically differentiated in terms of longer or shorter duration (i.e., the feature [±tense]). Therefore, we expect that Swiss German learners of French would assimilate the non-native feature [±voice] to the native [±tense] contrast, due to the great similarity and the functional equivalence of the two features; devoicing is predicted to occur more often in universally preferred positions such as the prepausal context. The corpus consists of 20 sentences (containing 6 voiced obstruents in 6 different phonotactic contexts), which were read by 10 high school students. An acoustic analysis permitted to categorize the 340 tokens into three discrete types: fully voiced, fully unvoiced, partially voiced. Chi-square tests yielded significant effects of the factors "context", "segment" and "speaker" on the variable "voicing". In particular, speakers pronounced 58% of the intervocalic obstruents as fully voiced, whereas they devoiced 85% of the prepausal tokens (thus, revealing both L1-based and universally preferred patterns).

This paper discusses the notion of "onsets" within the Onset Prominence representational environment (OP; e.g. Schwartz 2010, 2013), with empirical focus on the representation of glides and initial vowels. Both glides and initial vowels have been shown to exhibit ambiguous behavior across languages, which has been problematic for representational theories based on a linear string of segments. The OP environment is based on a hierarchy of phonetic events, incorporating structural ambiguities that may serve as parameter settings for the dual behavior of both glides and initial vowels. This approach eliminates the need for an ONSET constraint, and offers an explicit reference point by which the concept of markedness in prosodic structure may be defined.