EN
The present paper reports the results of an empirical study that was designed to provide acoustic evidence that there exist at least two different vowel reduction patterns in the Russian language. The acoustic characteristics of the three peripheral vowels [i, a, u] were examined. Given that low vowels are dispreferred in unaccented positions, particular attention was paid to immediately pre-tonic [a] sounds that result from 1st degree reduction and which are weakened to [e] by some speakers. The acoustic and statistical analyses revealed that vowel reduction in Russian is a speaker-specific phenomenon. Although all subjects centralised unaccented high vowels, two of them applied two degrees of reduction, whereas the other two exhibited a different reduction pattern as in their speech the differences between the acoustic parameters of the [a] and [e] sounds did not reach the level of statistical significance (p > .05). The acoustic data strongly suggest that if a speaker applies one degree of vowel reduction, then the [i, e, a, e, u] inventory is simplified to [i, e, u] rather than [i, a, u] as there is a highly significant difference (p < .001) between the sounds found in immediately pre-tonic position and the [a] sounds placed in stressed syllables.