Healthy social-emotional development in early childhood has lifelong implications for success in academic fields as well as in interpersonal relationships (Dunn & Brown, 1991; Goleman, 1995; Graziano et al., 2006; Huffman, Mehlinger, & Kervian, 2000; Keane & Calkins, 2004; Mayer & Salovey, 1997). At least one in five children under the age of six is diagnosed with emotional/behavioral problems in any given year, which unless addressed in early childhood, are likely to persist and require special education and/or mental health treatment (Campbell. 2002; Lavigne et al., 1996; Pianta & Ninietz, 1992). Psychodynamically grounded therapeutic early intervention is a language based approach which relies on building trusting teacher-child relationships and developing effective and affective teacher-child communication skills (Archer & Hosley, 1969; Hyson, 1994; Koplow, 1996; Kostelnik et al., 1998). Teachers in the therapeutic classroom use specific verbal strategies to guide children in being self-reflective and linking their affect with their behaviors (Ferber, 1996). This study examined the frequency of positive and negative teacher discourse and child responses as well as teacher-child dialogue patterns to determine whether designated categories of teacher discourse were likely to elicit specific categories of child responses across time. The results indicated that positive teacher discourse and positive child responses increased over time and that specific teacher discourse-child response sequences were significantly related. Positive teacher discourse elicited positive and some negative child responses, but negative teacher discourse most often elicited negative child responses, demonstrating that the words and language that teachers choose can directly impact the way in which a child will respond. Limitations of the study, as well as implications, and recommendations for practice are presented.