Paulina Zydorowicz

The study compares educated Poznań speech on the basis of a study byWitaszek-Samborska (1985, 1986) and a corpus compiled thirty years later. Thefeatures of Poznań speech, examined on 14 speakers from the corpus, include:voicing of obstruents before heterolexical sonorants (okszyg emocji), realizationof word-final ‹-ą› as [-ɔm] (idom tom drogom), realization of /stʂ tʂ dʐ/ as /ʂt͡ʂ t͡ʂ d͡ʐ/ (szczelać), the presence of the velar nasal [ŋ] before a heteromorphemic velarplosive /k/ (okienko), realization of word-final ‹-ej› as /-i(j)/ or /-ɨ(j)/ (lepi(j)), presenceof prothetic [w] before word-initial /ɔ/ (łojciec), presence of voiced /v/ in clusterswith preceding voiceless consonants (trwały), and realization of ‹-śmy› as [ʑmɨ](słyszelˈiśmy). The results suggest a change in Poznań speech and point towardsdialect levelling.

The goal of this paper is to investigate Polish phonotactics from the point of view of different measures of phonotactic preferability. The inventory of word-initial and -final clusters is extracted from a dictionary and analysed in accordance with two principles of phonotactic complexity, namely, the Sonority Sequencing Generalisation and Net Auditory Distance. Sonority entails measurements of distances between consonants expressed by the manner of articulation, whereas NAD uses the manner of articulation, place of articulation as well as the obstruent/sonorant distinction. These differences are likely to contribute to a different assessment of clusters, which is the main focus of this paper. Moreover, since a set of Polish clusters arise due to morphology, a distinction is drawn between phonotactic and morphonotactic clusters, i.e. phonologically and morphologically motivated. We are interested in verifying to what extent the principles under investigation reflect the relation between cluster preferability and morphological complexity. The analysis shows that NAD, as a more restrictive measure of phonotactics, rejects a larger portion of word-initial and -final clusters on well-formedness grounds. Secondly, we demonstrate that both principles generally show a strong relation between cluster preferability and morphological complexity.