The paper focuses on major structural differences between the language of heritage speakers of Russian living in the USA, the language of their parents (Emigre Russian), the standard language of the late twentieth century (CSR), and the spoken version of Russian used by large population groups in Russia and increasingly represented by the Russian media. The relationships between these four versions of Russian are complex and multi-dimensional; they may vary depending on the criteria of comparison. This variation is illustrated using two different structural characteristics, vowel reduction and analyticity in morphology.
After presenting several structural characteristics of the language of heritage speakers, the paper goes on to argue that the model of diglossia is most adequate and promising for the representation of the relationships between the Russian of heritage speakers and other versions of Russian. The diglossic representation has the following major implications: heritage Russian is a separate, self-contained linguistic system with its own set of rules; structural characteristics of heritage Russian have a bearing on linguistic competence, not just performance. Since heritage speakers do not have the knowledge of the linguistic system of standard Russian, they need to be presented with an explicit description of this system and cannot be expected to switch to it based on their own variety of Russian.