13.07.2015 Views

Untitled

Untitled

Untitled

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

Three-consonant sequences73voice and given the fact that [r] patterns with obstruents rather than withsonorants as far as the result of palatalisation is concerned, 20 it was suggestedby Scheer (2004) that trapped sonorants are, to use his term,‘demoted’ to regular obstruents (see also Rubach and Booij 1990a,b). One of the consequences of this observation is that trapped consonantsare not transparent to voice assimilation but are regular obstruentsand form an obstruent cluster with the neighbouring consonants. Inthe case of bó[p °r], for example, the final cluster behaves like an obstruentcluster in that both consonants undergo devoicing. The same holdstrue for word-internal trapped sonorants: CRCs form one single obstruentcluster, hence, its members must share the same voice specification,which is a general rule of Polish. A clear piece of evidence supporting theclaim that syllabic and trapped consonants must be assigned two differentrepresentations comes from Czech. On the one hand, both Polishand Czech exhibit obstruent word-final devoicing e.g. Polish chle[p] —chle[b]a ‘bread’, Czech holu[p] — holu[b]a ‘pigeon’. On the other hand, inCzech the TR clusters in the word-final position are not devoiced as theyare in Polish, compare Polish bó[p °r] vs. Czech bo[b®r]. Furthermore, wordinternallyCzech syllabic consonants are not transparent to voice assimilation,e.g. Czech [t®rvat], [k®rvE] vs. Polish [t °rfat¡‚], [k °rfi]. Thus, the conclusionthat can be drawn from the above data is that the transparency ofPolish trapped sonorants is not a consequence of the context alone astheir Czech syllabic cognates occur in exactly the same environment withoutshowing this behaviour. The explanation may be sought in the factthat in Polish certain sonorants behave like obstruents, while in Czechthey behave as vowels. This fact is further confirmed by the syllabic statusof such segments. The conclusion at which we arrive, therefore, isthat two separate representations are needed to distinguish syllabic fromtrapped sonorants. Let us now turn to another, no less relevant, piece ofevidence, namely stress placement. It is a generally accepted fact thatsyllabic consonants can be stressed. In Slovak, 21 for instance, stress alwaysfalls on the first syllable. If a word starts with a consonant clustercontaining a syllabic segment that consonant is stressed, e.g. [k®rmit] ‘feed’but if the sonorant is not syllabic, stress is placed on the following vowelas in [krava] ‘cow’. Similarly to Slovak, /r/ in Serbo-Croatian can alsobear primary stress. In the latter language stress falls on the only vowelin monosyllabic words. Note, however, that [t®rga] ‘square, gen.sg.’ has20It must be noted here that [r] and its palatalized versions [Z] and [S] are the mostcommon segments occurring in the trapped context.21The Slovak data have been collected from Blaho (2001), see also Rubach(1993).

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!