Phonology and Phonetics of Palatality Contrasts

Alexei Kochetov, University of Toronto

This paper provides an analysis of the contrasts involving palatalized and palatal coronal obstruents in the surface inventories of several Slavic languages. It argues that the observed variability of coronals is superficial and can be reduced to a limited number of phonological patterns conditioned by universal markedness and co-occurrence constraints on complex segment structures. The attested diversity of surface inventories is derived by language-specific contrastiveness requirements and enhancement effects. The account is presented in the framework of Optimality Theory with reference to phonological segment structure.

I propose that the presence or absence of palatalized or (pre-)palatal segments in an inventory can be captured by different rankings of constraints that maintain (PRES) and neutralize (NEUT) complex segments. Most Slavic languages (except Kashubian and Slovene) preserve the contrast. However, they differ in the surface realizations of the contrasting consonants. Thus, the inventories of voiceless coronal stops in Russian, Belorussian, Czech, and Polish are similar in the way that all of them have the dental and palatoalveolar affricates as well as the plain dental stop. However, they differ in what palatalized or palatal segment each inventory has. Thus, Russian has the palatalized /tj/, but lacks the palatalized dental affricate and (pre-)palatal segments. On the contrary, Czech has the palatal stop /c/, but does not make any use of other palatal or palatalized segments. The Belorussian dental /tsj/ and Polish prepalatal /cc/ are in a similar complementary distribution in the corresponding inventories.

I argue that the constraint on co-occurrence of palatalized and (pre-)palatal coronals (*tj-tsj-cc-c) follows from the underlying representation which is identical for these segments: a C-place-coronal specified for V-Place-coronal. Support for this representation comes from similar phonological patterning of these consonants in the analyzed languages. The proposed representation has a range of surface realizations differing in constriction location and stridency. I demonstrate that the attested surface realizations result from language-particular rankings of several universal constraints: PRES (preserve a contrast), DIST (maximize distinctiveness), and IDENT (do not alter the phonological input) (cf. Flemming 1995; Kirchner 1997, etc.). In order to maximize the plain-palatalized contrast (DIST1) a language may enhance the contrast by providing additional auditory cues: DIST2 (change of constriction location), DIST3 (stridency), etc. While the minimal contrast (t - tj) suffices for Russian, it is not salient enough for Polish, which employs several enhancement features resulting in (t - cc), as illustrated in (1).

r1c1 PresPal Dist1 Dist2 Dist3 Ident /tj/ NeutPal
t - tj *!
t - tsj *!
t - c *!
t - cc
t *!

I demonstrate that the use of enhancement by a language tends to correlate with the actual number of non-palatalized-palatalized contrasts in a given inventory and the degree of "automatic" palatalization in that language. Language-particular preferences for the contrastiveness of palatal or palatalized segments with other coronals also contribute to the surface inventory outputs. I show that the proposed analysis holds for South Slavic (Macedonian and Serbo-Croatian dialects), where underlying palatal consonants have a similar range of surface realizations.