Abstract
The study of heritage languages is opening new insights into the properties of language acquisition and attrition. While there is plain-text child and bilingual data (e.g., MacWhinney 2000, Tao&Ming 2008), there are no annotated corpora for the study of language attrition, acquisition and bilingualism. The presentation will describe such a corpus being built at Brandeis University. The corpus is based on unscripted speech from child, émigré, and heritage speakers (HSs) of Russian in the USA and Germany, and monolingual families in Russia. To date, we have 170 minutes of recording from the USA, 2122 minutes from Germany, and 435 minutes from Russia, and have transcribed an estimated 8900 words from the USA, 10500 words from Germany, and 12575 words from Russia (average 65 wpm). The presentation will discuss principles of participant selection, transcription of oral speech, working with language data from small children, and planned grammatical annotation of the corpus.