Binary Classification of Russian Nominal Declension Based on Sonority

Ronald F. Feldstein, Indiana University

The partial similarities of Russian declension types have led to great variations in the determination of how many subtypes should be recognized. Some systems recognize a four-way or five-way division (based on the set of forms found in all singular cases), while others recognize a three-way system, based on the three-way division of the nominative singular. Certain schemes exclude the notion of agreement gender, while the most traditional systems proceed from the notion of gender and often fail to clearly distinguish inflectional sets from genders. Ivic (1972, Sistema padezhnyx okonchanij sushchestvitel'nyx v S-Xom jazyke), devoted to Serbo-Croatian declension, proposed that the declension types should be considered part of a binary system, based on whether the nominative singular lacked an -a ending (Declension I) or possessed an -a (Declension II). Internal variation within these sets was taken care of by predictive rules, based on agreement gender.

My proposal takes Ivic's notion of a binary declensional division as a starting point, but establishes the binary division on the basis of a large number of binary sonority oppositions within the Russian case system. It turns out that the nature of these binary sonority oppositions is closely correlated with the semantic case divisions originally proposed by Roman Jakobson in his 1958 paper, "Morfologicheskie nabljudenija nad slavjanskim skloneniem." The singular and plural each have two blocks of cases, direct and oblique. Each such block possesses its primary case for sonority opposition (which happens to lack the Jakobsonian feature of pure directionality); the nominative occurs in the direct block and genitive in the oblique block. In both the singular and plural, these cases manifest a clear binary division, based on sonority. NSg, GSg, and NPl all oppose nouns with the maximally sonorant -a desinence to those which have a desinence of lesser sonority; the GPl, by contrast, opposes nouns with minimally sonorant zero desinence to those which have a mid-vowel desinence of greater sonority. The actual distribution of nouns is different for each of the four main case slots. A number of interesting conclusions can be reached by considering the various sets of binary sonority features which occur in the Russian nominal system. For example, a negative sonority specification in NSg, GSg, and NPl (e.g., tetrad') is virtually always related to the semantic property of non-human (unless the special suffix -er- occurs). If one includes all nouns in the Zaliznjak grammatical dictionary, there is a total of 13 patterns of binary features across the four major case slots referred to above, with implications for the prediction of gender.

The purely directional cases (accusative and dative) also have interesting binary properties, related to syncretism and inversely redundant to the sonority of the non-directional nominative and genitive. Thus, among the direct cases, a plus-sonority -a in NSg implies non-syncretism in the ASg (e.g., lampa), while minus-sonority in the NSg does imply ASg syncretism (stol). Likewise, among the oblique cases, a plus-sonority -a in GSg implies non-syncretism in the DSg (stol), while a GSg minus-sonority value (-i) does imply DSg syncretism. However, these are only the initial binary divisions that are based on sonority, i.e., the immediate constituents. It happens that some cases (i.e., the least marked NSg and ASg) can have secondary binary divisions, such as that of -a vs. non-a among the syncretic ASg nouns, related to animacy.

My proposal mainly concentrates on paradigmatic sonority oppositions of the various noun classes, but briefly touches on syntagmatic sonority contrast between the desinences of a noun and its modifier. Syntagmatic sonority agreement or disagreement has important implications for the gender system (cf. vrach voshel with agreeing minimal sonority (zero endings) vs. vrach voshla with disagreeing minimal/maximal sonority; also Ėta devushka with agreeing maximal sonority values vs. Ėtot voevoda with disagreeing sonority). Thus, the paper connects the use of sonority differences on both the paradigmatic and syntagmatic axes and explores the extent of sonority as a signalling device within Russian and other Slavic morphological systems.