The present paper tries to test the validity of the notion of congruence and the Uniform Structure Principle against an Algerian Arabic-French code-switching corpus. These two principles represent an important corner stone underlying the latest version of the Matrix Language Frame Model as it has been developed by Myers Scotton in 2002 and 2006. Data from naturally-occurring conversations is analysed within Myers-Scotton’s Matrix Language Frame model (1997, 2002, 2006 and 2015) and its supportive model -the 4-M model. Our investigation of bilingual language behaviour of Algerian speakers is conducted from a micro sociolinguistic perspective. The Matrix Language Frame Model’s descriptive paradigm succeeded to a large extent in analysing the different insertions in our corpus. To explain and interpret the observed instances of code switching we opted for Matrix Language Frame model’s explanatory principles (i.e. congruence and structural uniformity) particularly in the case of challenging code-switching instances. The results of our investigation highlight the fact that both congruence and the Uniform Structure Principle are flexible paradigms that can be adapted and interpreted to account for language-specific patterns and structures. Another important observation is the case of insufficient congruence between Algerian Arabic and French. Our corpus revealed two possibilities in this situation. The structural uniformity of both languages is preserved by using compromising strategies (Embedded Language islands or internal Embedded Language islands) which is in accordance with Myers-Scotton’s model. The other possibility that constitutes a challenge to the MLF model is the fact that Matrix Language structural uniformity is imposed across mixed constituents at the expense of Embedded Language structural uniformity.