The focus of this study is to use Electromagnetic Articulography (EMA) data to investigate the difference between word-initial and word-medial coordination, how coordination is affected by segment quality and stress, and to model coordination as it relates to syllabification. There are four key results of this study. First, segments in word-medial position have looser constrictions than segments in word-initial position, demonstrating that articulatory pressure is a driving factor in word-medial lenition. Second, word-medial syllables differ in coordination from word-initial sequences; however, current methods of analysis do not provide adequate tools to analyze such variability. Third, jaw oscillation does provide a metric for analyzing variability, crucially, demonstrating that word-medial syllables are more variable than initial syllables and unstressed medial syllables are more variable than stressed medial syllables. In addition, clusters coordinated within the same jaw phase show an increased likelihood to show c-centering and rightward shift. Thus, there is a gradient likelihood for medial clusters to syllabify according the Onset Maximization Principle, where initial stressed syllables are most likely and medial unstressed syllables are least likely, with the cluster syllabifying instead as a coda and onset. Fourth, coordination stability is influenced by by both coupling to jaw phase and intrinsic coordination between segments; both homorganic clusters and heterorganic clusters show coordinative stability, showing that jaw rather than gestural overlap between articulators determines stability; however, cluster composition and vowel quality interact and influence patterns of coordination such that a cluster is more likely to be syllabified as an onset when the cluster and vowel match in backness. Together, these findings demonstrate that word-medial coordination is dynamic and influenced by both top-down pressures, like stress, and bottom-up pressures, like segment quality. Furthermore, these findings demonstrate that jaw oscillation can be used as a diagnostic for motor planning and syllabification, providing a tool to improve existing models of planning and syllabification.