Two approaches to developing a theoretical platform for the study of sociosyntax are compared. The first approach adopts a sentence-based theory of syntax and an objectivist theory of semantics, which serves to qualify forms as variants of a sociolinguistic variable. The second approach replaces the sentence with the sign (signaled by grammatical formatives and word-order patterns) as the primary organizing principle of grammar and the basic carrier of meaning. It also replaces objectivist semantics with subjectivist-construal, functional-semiotic semantics under which grammatical meanings offer particular speaker perspectives. In this second approach the alternants of interest to sociolinguists are in almost every case different in their component meanings, yet amenable to conceptualization as variants of a sociolinguistic variable. Variation is seen as resulting from pressures exerted simultaneously by the exigencies of communication and the force of both internal and external sociolinguistic factors. This approach offers considerable advantages for the study of sociosyntax, advantages that pertain to both overall theoretical coherence and clearer paths to the explanation of the effects of predictor variables. We illustrate these advantages with analyses that demonstrate the impact of switch-reference on the presence and absence of Spanish yo, tú, ella, él, etc. ('I, you, she, he').

The study of syntactic variation has lagged behind the study of phonological variation in sociolinguistics, despite early claims that ‘[t]he extension of probabilistic considerations from phonology to syntax is not a conceptually difficult jump’ (Sankoff 1973: 58).1 Documented challenges to the study of syntactic variation include the increased difficulty in circumscribing a linguistic variable when dealing with levels of the grammar above phonology (Tagliamonte 2012: 206–7), and problems of convincingly quantifying syntactic variables which occur less frequently than phonological variables (Rickford et al. 1995: 106). Added to this, it has been argued that syntactic variables are less subject to social evaluation than phonological variables (Labov 1993, 2001: 28; Levon & Buchstaller 2015) and, even when they are, they tend to have ‘quite fixed social meanings associated with external facts like class and particularly education’ (Eckert 2018: 190).