For over four decades, International Teaching Assistants (ITAs) have faced criticism by undergraduate students for their linguistic and teaching abilities. One area of research on ITAs' linguistic abilities focused on their pronunciation and generally revealed non-standard patterns of segmental and suprasegmental features (Anderson-Hsieh et al., 1992; Kang, 2010) that can result in lower comprehensibility for ITA speech (Hahn, 2004; Kang et al., 2010). ITAs' discourse intonation (i.e., the role of intonation in natural interaction), and prominence (i.e., stress on a syllable to mark it as salient in a phrase or sentence) in particular, has been found to differ from native-speaker standards. Prominence plays an important role in signaling the structure of information. TAs who speak American English as their first language (ATAs) generally mark new information as prominent and given information as non-prominent (Pickering, 1999; Wennerstrom, 1997). ITAs, however, show irregular patterns of prominence marking in their classroom speech, making the processing of their speech more challenging for their students: They mark multiple words as prominent, pause in the middle of tone units, misplace prominence on given information, and miss prominence on new information (Gorsuch, 2011; Hahn, 2004; Pickering, 2018; Tyler et al., 1988). But research on native and non-native pronunciation has traditionally used elicited data consisting of words and sentences, rarely using extended discourse to study the interactions between pronunciation features and communicative functions of language. This dissertation has taken a register approach in developing a corpus of teaching assistant classroom speech, which is prosodically annotated to enable studies of the relationship between prosodic features such as prominence and discourse-level aspects of language such as information structure. This dissertation begins with an overview of ITAs' linguistic performance and communicative competence as a framework for ITA training, followed by a discussion of the corpus approach to studying prosody in classroom speech. The interactions between information structure and discourse intonation are covered in the second chapter using two models of discourse intonation. The dissertation then includes three papers. The first describes the steps involved in the design, compilation, and development of the Corpus of Teaching Assistant Classroom Speech (CoTACS). The creation of this corpus was motivated by the need for developing prosodically-annotated corpora of non-native speech in instructional settings, for use in both research and teaching. CoTACS, once publicly available, will enable more studies of ITA speech to identify prosodic characteristics of their speech that can impact undergraduate students' comprehension. It will also allow for more robust replications of previous studies that used small datasets, in order to validate previous findings. The first paper describes the register perspective taken in designing the corpus and analyzing the domain of instructional speech by TAs, followed by details of the sampling methods, speakers' demographics, data preparation, orthographic transcription, audio-transcript alignment, prosodic annotations, and addition of part-of-speech (POS) tags and metadata to the corpus files. The paper ends with a discussion of the composition of the corpus and plans for its availability. The second paper in this dissertation presents a proof-of-concept study of prominence placement in ITA and ATA speech. Using a sample of CoTACS data, this paper showcases how the corpus can provide data for research on prosodic features such as tone units, prominence, and pauses in speech by TAs from different linguistic and disciplinary backgrounds. Through a series of examples, this paper reports a qualitative analysis of speech from five ATAs and 10 ITAs. Findings document the patterns of similarities and differences between these two groups in their use of prominence to mark given and new information. The potential impact of these prosodic choices on information structure are discussed with respect to Brazil's (1997) model of intonation. The third paper introduces the potential of CoTACS to provide teachers and learners support for pronunciation instruction, using the findings from a small-scale study of prominence placement with data taken from six CoTACS speakers. The chapter also explores using corpus approaches to determine teaching targets for prosody, introducing activities for using CoTACS to teach intonational features such as prominence. The implications connect the use of CoTACS to teaching pronunciation to learners from a variety of L1 and disciplinary backgrounds. The dissertation ends with a discussion of implications for further research in second language pronunciation and corpus linguistics, as well as implications for teaching based on the results regarding prominence placement in ATA and ITA speech and using CoTACS data for developing teaching materials. [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.]