Studies adopting a Complex Dynamic Systems Theory framework have often fallen into one of two camps: those investigating variable linguistic subsystems (namely, CAF studies) and those concerned with non-linguistic variables displaying the characteristics of dynamic systems, with little concern for their connection to linguistic development. Examining asynchronous interactional data in an 8-week email exchange between a bilingual speaker of Spanish and English and an advanced Chinese learner of English, this study attempts to reconcile these two camps by exploring the extent to which syntactic complexity and article accuracy vary situationally as a function of topic and willingness to communicate (WTC) in L2 written performance over time and by investigating any potential relationships between complexity and accuracy. A qualitative, inductive analysis reveals that topic bears the characteristics of a dynamic system. The results also show that, for both the native speaker (NS) and non-native speaker (NNS), syntactic complexity, measured as subordinate clauses per T-unit and complex nominals per T-unit, varies situationally with topic, shifting not just over time but within a single email. As might be anticipated, article accuracy exhibits the typical instability of a dynamic system for the NNS but not for the NS, whose stable performance may reflect a permanent attractor state. Both the NNS' and NS' article accuracy are subject to cross-linguistic influence. Evolving correlations between complexity and accuracy suggest that certain topics and the WTC associated with them may induce trade-offs between the two subsystems.