Abstrakt: | The aim of this paper is to compare two analyses of Polish initial consonantal sequences. The starting point for the discussion is a brief presentation of the solutions available within the standard Government Phonology approach (henceforth GP). The findings are then confronted with the alternative analysis – a strict CV version of the standard GP. As the research unfolds, it becomes clear that a strict CV model is not only more elegant and more economical, but it also explains the facts better. The only problematic cases for this alternative approach, i.e. the so-called trapped sonorants (consonantal sonorants that appear in an onset between two consonants of lower so-nority, as in e.g. [krt[ań ‘larynx’), are dealt with in the second half of this paper. The author attempts at resolving the problematic character of trapped sonorants by repre-senting them as left-branching structures, i.e. as spreading of the sonorant onto a pre-ceding empty nuclear position, similarly to syllabic consonants. This solution seems promising as it explains such sequences without introducing additional machinery. The proposal gains an extra relevance, especially as standard GP seems to overlook the puzzle of trapped sonorants altogether. However, when it comes to the analysis proper, a competing solution becomes available. Although both of them can solve the problem, the acceptance of one over another has some further consequences. |