One of the early insights in sociolinguistic variation was Labov's (1966) distinction between the grammar, a largely automatic system which produces and perceives linguistic material, and another module, eventually named the sociolinguistic monitor (Labov 1993), which assigns prestige to linguistic forms and depends on attention to run optimally. Subsequent work in the third wave of variation (Eckert 2012) showed that speakers navigate sociolinguistic terrain much more complex than high vs. low prestige, while the grammar vs. monitor divide was challenged by sociophonetic research demonstrating the grammar's sensitivity to detailed contextual information, including social information (e.g. Strand 1999).
I argue that while the sociolinguistic monitor itself is insufficient for modeling sociolinguistic cognition, the core insight of multiple systems remains key. Instead of sociolinguistic-specific architecture, systems already posited to explain social and linguistic behavior can account for sociolinguistic behavior, with appropriate expansions and connections. I will discuss evidence for dissociations between systems related to indexical meaning, namely person perception, sociolinguistic production, linguistic perception and metalinguistic beliefs, as well as a possible difference between the person perception system and online sociolinguistic monitoring. I conclude with some next steps for understanding these systems.