Abstract:

Many current educational studies focus on identifying factors that contribute to successful learning. The researchers describe the behavior of effective teachers, including their skills, teaching methods, and learning styles. At the same time, since speech is a key tool for learning, the linguistic features of teachers’ speech should also be considered as a possible factor influencing the process of learning and the interaction between teachers and students.
Studying linguistic features of speech requires representative speech data. When describing a language, speech corpora are widely used. These corpora usually consist of substantial amounts of oral or written texts accompanied by a multi-level annotation. A corpus of teachers’ speech with linguistic annotation will allow further comparative research of more and less effective teaching practices and will serve as a source of examples of effective teachers’ speech. In this paper, we aim to outline the principles of constructing a corpus of effective Russian-speaking teachers.
The speech material for our corpus includes video and audio recordings from real school lessons. Currently, more than 70 lessons (around 40 minutes each) taught by effective teachers have been recorded and transcribed. We plan to include the comparative data from ineffective teachers in our corpus as well.
All files are provided with metadata: gender and age of a teacher, region where a school is located, subject, grade, and lesson stage (revising the material, explanation of new material, giving homework assignments, etc.). The assessment of the teacher's effectiveness (distinguishing between effective and ineffective teaching practices) is an essential part of this annotation. The principles for determining teacher’s effectiveness are described in (Sergomanov et al., 2023).
The linguistic annotation of the corpus starts with a multi-level classification of pauses that divide the speech into convenient segments for further analysis (Vinogradova et al., 2023). The preliminary analysis of audio recordings from two teachers revealed the presence of specific rhythmic pauses in their speech, presumably used to direct students' attention to important information. Both teachers prefer unfilled pauses to filled ones; the pauses between clauses are significantly more frequent than the pauses within clauses. According to these first results, we can hypothesize that the speech of effective teachers is closer to prepared monologues than to spontaneous speech.
The analysis of orthographic transcripts of the lesson allowed us to identify the most frequent collocations in the speech of effective teachers and to determine how the effective teachers use particular constructions and words. The effective teachers tend to use the pronoun “we” in the inclusive function (‘you and me’), but there are specific functions for certain subjects (e.g. history) as well. We are currently developing the principles of phonetic transcription of the recordings to obtain the information about tempo, phonetic reductions, etc.
During the conference, we are going to present a draft of our corpus and discuss the preliminary results of the linguistic analysis of the corpus material.
"The work is supported by Saint Petersburg State University (project #101747352 “The Role of Linguistic Features of Speech in Effective Teaching Practices: A Corpus-based and Psycholinguistic Study”) and “SberObrazovanie” (contract # 230712-107-ЮЛ)".

Keywords:

Spoken corpora, teachers’ speech, linguistic annotation, interdisciplinary approach.