Solving Russian velars: Palatalization, the lexicon and gradient contrast utilizationdoi:10.18710/1J0YZGDataverseNO2014-12-041Parker, Jeff, 2014, "Solving Russian velars: Palatalization, the lexicon and gradient contrast utilization", https://doi.org/10.18710/1J0YZG, DataverseNO, V1Solving Russian velars: Palatalization, the lexicon and gradient contrast utilizationdoi:10.18710/1J0YZGParker, JeffThe Ohio State University2014DataverseNOThe Tromsø Repository of Language and Linguistics (TROLLing)Parker, Jeff2014-12-032014Arts and HumanitiesRussianvelarspalatalizationcontrastphonologygradienceField: PhonologyTopic: consonantsTime-depth: synchronicThis dataset consists of (1) an excel file with type and token counts of all paired consonants word-finally and before non-front vowels, their probabilities, and the entropies of the pairs in each context; (2) the same entropies in separate files for word-final and before non-front vowels; and (3) R code to generate plots and perform statistical analysis. Article abstract: Palatalized velars in Russian are often considered exceptional because they are neither fully predictable, nor clearly unpredictable. They are an example of a common phonological relationship in which sounds have the potential to distinguish words but are only utilized in limited contexts and/or lexical items. These 'intermediate phonological relationships' (Goldsmith 1995) are problematic for traditional phonological theories which make a binary distinction between predictable sounds (allophones; dealt with in the grammar) and unpredictable sounds (phonemes; dealt with in the lexicon). To deal with intermediate phonological relationships in a principled way we must reconsider assumptions about the type and amount of information stored in the lexicon. In this paper I show that in Russian, both palatalized and non-palatalized velars occur in a variety of contexts, evidence that they have the potential to distinguish words. I also show, using information-theoretic metrics, that the potential is utilized to a minimal degree across both lexical items and phonetic contexts. However, and importantly, I show that many other consonants likewise do not fully utilize the (same) palatalization contrast across contexts. This suggests that velars are not an 'exception'; instead, they represent a relationship which lies at one end of a continuum along which the palatalization contrast is utilized. I argue that it is not velars, or intermediate phonological relationship
s more generally, that are at problematic. Rather, it is our assumptions about the type and amount of information speakers store that is at issue. I argue that memory-rich models of the lexicon, which assume a great deal of storage of phonetic, contextual and distributional information, better account for velars in Russian. Moreover, the type of relationship that velars represent is a natural and expected outcome in such models. Thus, Russian velars provide important evidence that pushes us to reconsider some of the basic assumptions of our phonological models and phonological relationships more generally, and the problem that has long vexed Slavists can be solved within a memory-rich model of the lexicon.United StatescorpusParker, Jeff. ‘SOLVING RUSSIAN VELARS: PALATALIZATION, THE LEXICON AND GRADIENT CONTRAST UTILIZATION’. The Slavic and East European Journal, vol. 59, no. 1, 2015, pp. 70–90. https://www.jstor.org/stable/44739582https://www.jstor.org/stable/44739582Parker, Jeff. ‘SOLVING RUSSIAN VELARS: PALATALIZATION, THE LEXICON AND GRADIENT CONTRAST UTILIZATION’. The Slavic and East European Journal, vol. 59, no. 1, 2015, pp. 70–90. https://www.jstor.org/stable/44739582Description of files.txttext/plain; charset=US-ASCIIentropies_preV.txtdata for scripttext/plain; charset=US-ASCIIentropies_wf.txtdata for scripttext/plain; charset=US-ASCIIparied_consonants_type_token_prob_entropy.xlsThis contains type and token counts of all paired Russian consonants before non-front vowels and word-finally; also includes probabilities and entropies calculated from the type and token countsapplication/vnd.ms-excelparied_consonants_type_token_prob_entropy_preV.txtPlain text version of tab 2 in excel filetext/plain; charset=US-ASCIIparied_consonants_type_token_prob_entropy_wf.txtPlain text version of tab 1 in excel filetext/plain; charset=US-ASCIIscript_graphs_and_stats.txtscript for data representation and statistical tests in Rtext/plain; charset=US-ASCII