This study provides an account for a long-term selective loss of L1 (Russian) morpho-syntactic and content components in early immigrants to the U.S. The analysis of naturally occurring data is carried out from the perspective of two theoretical approaches--three models developed within language contact (Myers-Scotton 2002, 2005) and the Activation Threshold hypothesis as a component of a neurolinguistic approach to bilingualism (Paradis, 2004, 2007). The results show that the language contact approach is useful in identifying morpheme types that are most vulnerable to attrition. The second approach helps explain the differential rate of loss of content morphemes in a variety of topics and account for variability in the rate of attrition of late system morphemes through frequency factors. The study demonstrates that by crossing the boundaries of one theory and one view of language researchers can achieve a stronger explanatory power and identify the common and complementary features that both models provide.