CALLFRIEND Russian Text (LDC2023T09) was developed by the Linguistic Data Consortium and consists of transcripts for approximately 48 hours of telephone conversations (100 recordings) between native Russian speakers. The calls were recorded in 1999 as part of the CALLFRIEND collection, a project designed primarily to support research in automatic language identification. One hundred native Russian speakers living the the continental United States made a single phone call, lasting up to 30 minutes, to a family member or friend living in the United States. Corresponding speech data is available as CALLFRIEND Russian Speech (LDC2023S08). The transcripts and lexicon are presented as plain-text, tab-delimited files with UTF-8 character encoding. In the transcript .txt files, the four main fields on each line (begin_offset, end_offset, speaker_label, transcript_text) are separated by tabs. Each contains a list of time-stamped segments in order according to their begin_offset values, with no blank lines. The lexicon covers the word forms in the 97 transcript files. The main lexicon table contains three columns per row: Cyrillic orthography, as found in one or more transcripts Phonetic transliteration Numeric representation of syllabic stress There is also a phonetic map file and a pronunciation rules file. The phonetic transliterations in the main lexicon table were derived programmatically based on information in those two files. The phonetic map file contains Cyrillic letters and their corresponding ASCII/Latin letters to represent the phonetic values. The pronunciation rules file is a list of "rewrite rules", presented as regular expressions, to implement a small number of context-dependent changes in the pronunciations of certain letters.