![Sample our Behavioral Sciences journals, sign in here to start your access, latest two full volumes FREE to you for 14 days]({"type":"image" , "src" : "/sda/110801/TOC-Subject-Sample-2021-Behavioral-Sciences.jpg"})
Abstract
The main goal of this research was to examine how readers of Russian assign stress to disyllabic words. In particular, we tested the claim that the process of stress assignment in Russian can only be accomplished lexically. Eleven potential non-lexical sources of evidence for stress in Russian were examined in regression and factorial studies. In Study 1, onset complexity, coda complexity, the orthography of the first syllable (CVC1), of the second syllable (CVC2), and of the ending of the second syllable (VC2) were found to be probabilistically associated with stress in Russian disyllables. In Studies 2 and 3, it was shown that Russian speakers do use 3 of these cues (CVC1, CVC2, and VC2) when making stress-assignment decisions. These results provide evidence against the idea that the nature of stress in the Russian language is so unpredictable that stress assignment can only be accomplished lexically. These results also suggest that any successful model of stress assignment in Russian needs to contain mechanisms allowing these 3 orthographic cues to play a role in the stress-assignment process.
Notes
1 Here and in all subsequent examples, Roman transliterations of any Russian words are given in square brackets.
2 Because there is no available electronic database that contains relevant information about various lexical characteristics of Russian words, we had to check those characteristics manually for each word. Therefore, for practical purposes, only words with a frequency of more than 1 per million words were included in the database created for this study.
3 The VC1 and VC2 components, as defined in the present work (i.e., components that comprise the syllable's vowel and the following consonants), are referred to in the previous literature as rimes (Treiman & Kessler, Citation1995).
4 Therefore, a word in a completely trochaic neighbourhood would get a 1.0 trochaic stress consistency value, while a word in a completely iambic neighbourhood would get a 0.0 trochaic stress consistency value.
5 In calculating spelling-to-stress consistency measures, most researchers consider only words of the same syllabic length (Arciuli & Cupples, Citation2006; Arciuli, Monaghan, & Seva, Citation2010), although in some work in Italian all words of the language regardless of their number of syllables were taken into consideration during the computation of consistency measures (Burani & Arduino, Citation2004). No empirical investigation has been conducted to determine which approach provides a better reflection of the processes that take place during lexical stress assignment in word reading. We decided to use the former way of calculating spelling-to-stress consistencies as this approach appears to be more consistent with the architecture of the CDP++ (Perry et al., Citation2010), a model that proposes that “the processing system has information about the number of syllables of an orthographic input before stress pattern information is computed (non-lexically).”
6 It should be noted that both the type and token comparisons have limitations due to the fact that only words with a frequency of more than 1 per million words were included in the database used in the regression analyses. Potentially, this action may have affected the type analysis more than the token analysis because each word in the neighbourhood is assumed to be equally important in any type analysis, whereas words with very low frequencies will have very little impact in a token analysis. Nonetheless, as will be seen, the type analysis appeared to be the more reliable analysis in our studies.
7 Recall that trochaic stress was coded as “0” and iambic stress was coded as “1”. Therefore, a negative z value in this analysis denote a significant association between a predictor variable and trochaic stress, while a positive z value represent a significant relationship between a predictor variable and iambic stress.
8 In this research, morphology as a potential stress cue was not examined directly. In Russian, there are a few derivational morphemes, mainly of foreign origin (e.g., -изм (-ism) as in фашизм (fascism), атеизм (atheism)) that may be predictive of assigned stress patterns; however, the majority of derivational and inflectional morphemes are not associated with a single stress pattern and, thus, are likely not highly reliable sources of evidence for stress.
9 As noted, Russian is transparent in its mapping of orthography to segmental phonology, but quite opaque in its lexical stress system. This characteristic of Russian was apparent in the distribution of types of errors in Study 2. Overall, the Study 2 error rate in word naming was 9.72%, with 94.13% of these errors being stress-assignment errors. The initial analysis reported for Study 2 (the analysis of stress patterns assigned by readers) was essentially an analysis of stress-assignment errors. No analysis of the other types of errors was undertaken due to their extremely low number.
10 Although VC1 contributed significantly to the final model in the type consistency analysis of Study 2, because it was not significant in the token consistency analysis in Study 2 nor did it emerge as a significant predictor of stress in the type consistency analysis in Study 1, the argument that it might be a viable cue to stress assignment in Russian is weak. Therefore, VC1 consistency was not one of the variables investigated in Study 3.
11 Similar to Study 2, errors in Study 3 were mainly stress assignment errors. Overall, the Study 3 error rate in word naming was 5.33% with 97.55% of these errors being stress-assignment errors.