Untitled

102 Bogus clusters, syllabic consonants and vowel syncope. . .German. This choice is justified by the fact that both languages, similarlyto Slavic, abound in syllabic consonants. Moreover, the phenomenon inquestion is very productive and well-documented.The most evident and at the same time the most general observationconcerning the consonantal inventory of English and German is the factthat some of the consonants can play a syllabic role. In other words, suchconsonants take over the syllabic duties. Consonants which are able tofunction in the way described above are generally referred to as sonorants.It follows that obstruents can never appear in syllabic clothes, atleast in the Indo-European languages. 1The last few decades witnessed a dramatic shift in the representationof syllabic consonants. In the SPE tradition syllabic consonants weredescribed as consonants which possessed a [+syllabic] feature specification.Together with the development of non-linear frameworks syllabicconsonants were perceived as segments which can change the constituentaffiliation. More specifically, sonorants were ascribed a special abilityto leave the consonantal position and move to the vocalic one. Thechange of place, however, was possible on condition that the precedingvowel (usually the schwa) had disappeared. In more recent frameworkssyllabic consonants are perceived as special in that they are doubly linked.This simply means that while being linked to a consonantal slota sonorant in certain cases can additionally dock on to the nuclear position.As mentioned in the previous chapter (section 3.5) the double-linkingrepresentation of syllabic consonants is especially true in the GovernmentPhonology framework. In the latter model there are at leasttwo theory-internal reasons which call for the representation mentionedabove. Firstly, in the Element Theory the realisation of an element dependson the position it occupies in the syllabic structure. This is clearlyobservable on the example of the element (I), which is realised either asthe vowel [i] or the semivowel [j]. The former segment can appear onlyunder a nuclear slot, while the latter one can be linked to a consonantalposition. It follows that if a sonorant were able to leave its original consonantalposition and dock on to the nuclear one, we would witnessa change in the realisation of the sonorant, just like in the [i] and [j] case.Secondly, the change of the constituent affiliation would imply resyllabification,which is banned anyway from the GP framework (1).1Piotr Ruszkiewicz pointed out to me that this statement may be too strong, aseven in English we encounter a few forms with a syllabic obstruent, e.g. psst [psÉt].Although rare, languages with ‘regular’ syllabic obstruents are reported to exist, forexample, American native Northwest languages (see Czaykowska-Higgins andWillett 1997, among others).