To date, few studies have investigated the production of linguistic complexity and accuracy with naturalistic data. Very often, the data are obtained in a laboratory or laboratory-style settings, with tasks defined by the researcher rather than the instructor (Hatch, 1978; Seedhouse, 2004). Additionally, replicability of studies that investigate the production of complexity, accuracy, and/or fluency (CAF) has been made difficult by the myriad ways that researchers have operationalized the constructs in their research (c.f. Ellis & Barkhuizen, 2005; Housen & Kuiken, 2009; Housen, Kuiken, & Vedder, 2012; Norris & Ortega, 2009; Pallotti, 2009). Further, few studies use a Dynamic Systems Theory lens when researching CAF and the development of CAF. The current longitudinal case study investigates how task affects the production of linguistic complexity and accuracy by three intermediate students of Spanish using data collected in the participants' regular classrooms over one academic year. A DST framework is used to reflect upon each student's developmental trajectory over the course of the study. The data were transcribed and separated into AS-units, and then further coded using two global syntactic complexity measures and one specific syntactic complexity measure, and one global accuracy measure as well as one specific accuracy measure. In order to determine task effects, ANOVAs were performed on each student's data. A multi-level mixed effects model was used to determine whether there were any interactions between time and task type. Results of the ANOVAs showed that task affects each student's production in a slightly different way, while the multi-level mixed effects modeling showed that verbal accuracy alone showed an interaction between time and task type. Results of the longitudinal analysis of the oral production of linguistic complexity and accuracy using a DST lens showed that the students' production did vary over time and that each student followed her or his own trajectory over the course of an academic year. These results also showed that there were some trade-off effects with the measures of linguistic complexity and accuracy, in that when complexity measures increased, there was a tendency for accuracy measures to decrease. [The dissertation citations contained here are published with the permission of ProQuest LLC. Further reproduction is prohibited without permission. Copies of dissertations may be obtained by Telephone (800) 1-800-521-0600. Web page: http://bibliotheek.ehb.be:2222/en-US/products/dissertations/individuals.shtml.]