In order to support instructors who are non-native speakers of English (NNS) in preparing for delivering lectures in an English-medium science and engineering program, we started the OnCAL (Online Corpus of Academic Lectures) project in 2010. Transcriptions of lectures delivered mostly by native speakers of English (NS) were obtained from MIT Open Courseware (MIT OCW) and Stanford Engineering Everywhere (SEE), and an online corpus was built. As of September 2015, the corpus comprises 430 lecture transcriptions, which contain 3.5 million words. This corresponds to about 395 hours of recorded lecture time. The lectures are from courses that cover a wide range of fields in science and engineering: biology, chemistry, physics, mathematics, and computer science as undergraduate courses, and advanced mathematics, robotics, and artificial intelligence as graduate courses. Pedagogical functions have been identified in the teacher utterances to allow OnCAL users to find examples of how teachers use language to realize specific pedagogical purposes. The examples of utterances found through OnCAL can thus be seen by NNS teachers as a reference for their own lectures. In the present work, we analyze how mathematical equations are explained in lectures. In this type of explanations teachers make use of and are aided by visual modes, but we believe that the spoken language that accompanies the other modes still have to be very accurate.