This paper proposed a new phonological framework in order to describe "intonational neutralization" in Russian. The first half of the paper described Russian intonation patterns in a framework with a laconic representation using an autosegmental approach (e.g. Pierrehumbert 1980). The latter half of the paper explained a mechanism of so-called "intonational neutralization" in this framework that has not been fully explained in other frameworks.
"Intonational neutralization" is a phenomenon where both a rising-low and a rising-high intonation patterns become a rising pattern when they appear at the end of a phrase and lose their contrast. Although this phenomenon has been described in various ways (e.g., using graphics written by impressionistic lines or complicated parametric signs), none of them has satisfactorily explained what causes "neutralization".
In the current proposed framework, intonation patterns were described as a string of two phonological tones, high (H) and low (L), both of which were determined by a relative height of pitch. Each phrase consisted of a "pitch accent" that was realized on a stressed syllable and a "phrase tone" that characterized a pitch movement from the pitch accent to the end of the phrase. Four types of pitch accents (H+L*, H*+L, L+H* and L*+H) and two types of phrase tones (H and L) were proposed.
In this framework, the mechanism of "intonational neutralization" was explained as follows. Two different underlying intonation patterns L+H* L and L+H* H were realized as L+H* on surface when the last syllable of the phrase was a stressed syllable that was associated with a pitch accent (L+H*). Since no syllable that followed the stressed syllable was observed in the phrase, a phrase tone (L or H, respectively) was not realized. This paper interpreted this phenomenon as "truncation", a phenomenon that has been observed in other languages (see e.g. Ladd 1996).