Untitled

Syllabic consonants103(1) Projection Principle (Kaye et al. 1990:221)‘Governing relations are defined at the level of lexical representationand remain constant throughout a phonological derivation’The Principle in (1) excludes any changes in governing relations duringthe course of derivation. In other words, resyllabification is prohibitedand a melodic unit that is linked to a consonantal position cannot surfaceunder a nuclear slot. Thus, it follows that the optimal representationfor syllabic consonants is the one where the sonorant is linked to a consonantalslot, while at the same time it spreads to a neighbouring nuclearposition. The idea is further discussed in the immediately following subsection.2.1. Some English factsThe general observation emerging from the short introduction above isthat in English, just as in several other languages, sonorants are grantedthe right to dock on to a nuclear position. 2 Moreover, in order to playa syllabic role a sonorant spreads to the left, i.e. to the preceding nucleus.What is crucial, however, is the fact that this analysis requiresa vowel to step aside and make room for the following sonorant. In ordernot to anticipate the discussion which will appear later in this chapter,we only note here that the schwa for one reason or another becomesdelinked from the nuclear position. In this way a receding schwa enablesa neighbouring sonorant to take over its duties. However, it is not truethat all sonorants have an equal opportunity to become syllabic. Thus, inEnglish only nasals, the lateral and the post-alveolar approximant canplay the syllabic role. Furthermore, the syllabicity of the velar nasal ismarginal simply because this nasal never appears after schwa (see Szigetvári1999, Gussmann 1998). Thus, every occurrence of the syllabicvelar nasal is the result of the process of place assimilation, e.g.chicken [tSIk«n] > [tSIk ®n] > [tSIk®N]. Finally, it should be clarified here thatin the majority of cases the phenomenon in question depends on the tempo2One may want to include here the forms where the schwa is lost between twoobstruents, e.g. difficult [ ® dIfIk®lt] > [ ® dIfk®lt], potato [p« ® teIt«U] > [p ® teIt«U], etc. Since, however,the clusters in question do not contain a sonorant they must be recognised as theinstantiation of a different phenomenon, see Abercrombie (1967) and Rodgers(1998).