BackHome
Refresh: Off
Paper Summary
Share...

Direct link:

Teachers’ Enthusiasm and their Natural Language in Class – a Sentiment Analysis Approach

Fri, April 25, 9:50 to 11:20am MDT (9:50 to 11:20am MDT), The Colorado Convention Center, Floor: Terrace Level, Bluebird Ballroom Room 3C

Abstract

Objectives and Theoretical Framework. There is compelling evidence that more enthusiastic teachers show better teaching quality (Fauth et al., 2019; Lazarides et al., 2021). However, to date, little is known about what it is that enthusiastic teachers actually do differently than less enthusiastic teachers to produce high-quality instruction, and theoretically well-founded explanations and empirical evidence to this question are rare (Keller et al., 2016). In the present contribution, we shed light on one specific potential mechanism through which teacher enthusiasm may produce high-quality instruction, namely the sentiment of the teachers’ language. We propose that high teaching enthusiasm manifests itself in teachers’ use of verbally positive expressions in the classroom.

Methods and Data Sources. The data reported herein stem from a larger longitudinal field study conducted at German secondary schools, tracking mathematics teachers and one of their classes for one school year, involving four self-report surveys, plus one video assessment administered past mid-term. The video recordings were complemented by short post-lesson diaries directly after the videotaped for teachers and students. The resulting final sample for the present analyses consisted of N=19 teachers (n=10/9 female/male) and their N=393 students (n=204/143 female/male, n = 46 not specified). Teachers’ habitual enthusiasm was assessed at the T2 survey administered 4 months ahead of the videotaped lesson (5/4 items; Omega=.87/.76 for teachers/students). Additionally, situation-specific teacher enthusiasm was assessed through two items each from the student- and teacher perspective. We realized the sentiment analysis using Remus et al.’s (2010) emotion lexicon SentiWS (v2.0, 2019). We obtained three sentiment indexes; an overall sentiment index averaging all scores for the Senti-WS-detected sentiment-carrying words within each teacher’s transcript, and a positive/negative sentiment index by averaging only across the positively/negatively valenced words.

Results and Scientific Significance. Table 1 shows the results for the teacher enthusiasm—sentiment associations (Kendall’s tau controlling for wordcount). There were significant, medium-sized positive links between teachers’ self-reported enthusiasm and their overall and positive (but not negative) verbal sentiment as expressed during the videotaped lesson. Correlations were larger for the post-lesson diary ratings than for habitual ratings. Correlations with student ratings were smaller and did not reach statistical significance.
This is the first study to employ sentiment analysis on transcripts of German teachers’ in-class talk. Our results imply that teacher enthusiasm drives the valence intensity of positively connoted words in the classroom, while it does not dampen the negative valence of teachers’ negatively connoted word use. The lower correlations with the student ratings may be due to students being less valid and suitable sources of inquiry for assessing teacher enthusiasm (Fauth et al., 2020; Vazire, 2010). Alternatively, teachers’ sentiment in their language may reflect attempts to “send” messages about the value and enjoyability of mathematics, which however are not necessarily “received” to the same degree by their students. The present study’s body of words, alongside the teacher- and student enthusiasm ratings, could be a starting point for further sentiment algorithm development, combining lexicon-based and machine-learning approaches resulting in optimized, pretrained sentiment algorithms to classify teacher speech.

Authors