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.

64 The phonological nature of the beginning of the wordthe one between the first two consonants, would have to be vocalised.This is not the case as the first nucleus remains phonetically silent, whichmay suggest that both forms have two different structures (22).(22)a. PG b. PG(C V) C|mV|PC|gV|eC|wV|P(C V)C|mV|PC|gV→IGC|wV|aFurthermore, note that in (22a) the vowel alternating with zero is allowedto properly govern a preceding nuclear position. This situation may seemproblematic at first sight as Polish alternating vowels are not ‘sound’ governors(Scheer 2004). This is clearly observable in the case of pies ‘dog’and the two diminutive forms piesek, pieseczek ‘dog dim./double dim.’ Ifalternating vowels were allowed to properly govern, we would arrive at *psek,*piesczek, which are ungrammatical. 14 Note, however, that both forms, i.e.pche³ and mgie³ would pose a problem if the first vowel were an alternatingone itself. Since the initial nucleus is lexically empty, it can be governed byan alternating vowel (see 29a below). In short, Polish alternating vowelscannot govern other alternating vowels but are able to govern lexicallyempty nuclear positions. 15 To make matters worse, we have assumed thatalternating vowels as lexically present can be silenced only by Proper Government(see section 2). If this assumption is correct, the representationin (22b) is ill-formed. In other words, the forms in (22) behave oddly becausewhat is the IG domain in one case (22b) is broken up by an alternatingvowel in another one (22a). Recall that we have already encountered a similarproblem while discussing the case of gra — gier ‘play, nom./gen.pl.’(see section 2.2). Moreover, both mg³a and pch³a are peculiar in anotherrespect, namely, when prefixed the former one behaves regularly and vocalisesthe prefix vowel, e.g. odemgliæ ‘defog’, while the latter one does not,e.g. odpchliæ ‘deflea’, instead of the predicted form *odepchliæ. 1614This constraint is in the spirit of the earlier accounts known as Lower (Gussmann1980a, Rubach 1984, Spencer 1986, Piotrowski, Roca and Spencer1992, Szpyra 1992a, among others), which states that jer surfacing is normallyconditioned by the presence of another jer in the immediately following syllable.15Note that this constraint does not hold in the aforementioned Derived Imperfectiveforms, where the stem vowel must properly govern the one in the prefix, e.g. zbieraævs. zebraæ ‘collect, ipf./pf.’16In order to arrive at the correct form we are again forced to resort to bracketingand domain structure [odP[pxlit¡‚]] (recall the discussion in section 2.2).

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

Saved successfully!

Ooh no, something went wrong!