Investigating Instructor Talk among Graduate Teaching Assistants in Undergraduate Biology Laboratory Classrooms

INTRODUCTION

Diverse groups of people are more effective at solving complex problems than homogenous groups of high-achieving individuals (Hong and Page, 2004). As such, if we want to address our most complex biological problems, we need diverse groups of scientists working together. Unfortunately, the sciences are particularly exclusive disciplines. Undergraduate students leave the sciences at higher rates than other disciplines, and who leaves can be predicted based on personal characteristics such as race, ethnicity, gender, and sexual orientation (Seymour and Hewitt, 1997; Hughes, 2018; Seymour and Hunter, 2019). The findings of student-centered research suggest turning our focus to the choices, behaviors, and approaches of instructors to improve undergraduate science education and increase diversity and inclusion in the sciences.

Instructors shape the classroom environment through what they say and do. Instructor Talk, referred to as noncontent language, is “any language used by an instructor that is not directly related to the concepts under study but instead focuses on creating the learning environment” (Seidel et al., 2015; Harrison et al., 2019). Here are two examples of previously recorded instances of Instructor Talk (Harrison et al., 2019) that directly reference student learning in contrasting ways:

1. “If you couldn’t remember at the end of the class what you learned in the beginning, then the learning is really quite useless, right? Because we really hope that you’ll remember 5 years from now.”

2. “So make sure you understand this. It’s going to be very, very valuable for scoring high points there.”

The first example of Instructor Talk encourages students to learn for the long-term, while the second example encourages students to learn for points. Such contrasting examples illustrate how aspects of noncontent instructor language can shape the classroom learning environment.

Based on these characteristics of noncontent language, we hypothesize that Instructor Talk may have a profound impact on student learning, inclusion, and success in biology. Instructor Talk research thus far has been informed by the theoretical and phenomenological frameworks of instructor immediacy, student resistance to active learning, and stereotype threat. First, instructor immediacy—the perceived sociopsychological distance between instructors and their students—can be shaped by an instructor’s language and behaviors to affect student learning and sense of inclusion (Mehrabian, 1971; Kearney et al., 1988; Kelley and Gorham, 1988). Second, instructors may worry that students will resist innovative teaching techniques; however, the way instructors prime their students for active learning could greatly influence the chances of student resistance (Seidel and Tanner, 2013). Finally, a phenomenon known as stereotype threat—when people are concerned with confirming negative stereotypes about groups to which they belong—can negatively impact students’ performance on high-stakes exams (Steele and Aronson, 1995). Instructors could impact student performance by either inducing or mitigating stereotype threat with their noncontent language. These factors—instructor immediacy, student resistance, and stereotype threat—suggest that Instructor Talk may be central to student learning, performance, and sense of belonging in a science classroom.

Instructor Talk has been characterized in lecture classrooms taught by faculty at more than a dozen institutions, including faculty at 4-year universities and 2-year community colleges (Seidel et al., 2015; Harrison et al., 2019; Lane et al., 2021; Meaders et al., 2021), and postdoctoral scientists teaching online (Seah et al., 2021). Two frameworks to categorize Instructor Talk emerged from Harrison et al. (2019). These frameworks consisted of Positively Phrased Instructor Talk with five categories and 18 subcategories and Negatively Phrased Instructor Talk with five categories and 15 subcategories. Sampling from a large number and range of biology lecture courses taught by faculty, Harrison et al. (2019) categorized approximately 90% of Instructor Talk instances in the Positively Phrased framework. The remaining 10% was categorized with the Negatively Phrased framework, which included Instructor Talk that may discourage students or distract from the learning process. Notably, all instructors studied to date had participated in pedagogical training on evidence-based teaching. Therefore, the Instructor Talk characterized in previous studies may not reflect the kinds of Instructor Talk used by instructors with little to no pedagogical training and experience. One may wonder about the characteristics of noncontent language among instructors without pedagogical training or at earlier stages of their careers, like graduate student instructors.

While previous studies have characterized Instructor Talk in lecture classrooms taught by faculty, we have yet to characterize Instructor Talk in laboratory courses taught by graduate teaching assistants, hereafter referred to as GTAs. Graduate students play a key role as instructors. Importantly, science, technology, engineering, and mathematics (STEM) graduate students teach more than 90% of laboratory sections at research institutions in the United States (Sundberg et al., 2005). STEM graduate students also come from more diverse backgrounds than faculty (National Science Foundation, 2021). Interestingly, a congruence between the gender or racial and ethnic identities of GTAs and undergraduates does not appear to correspond to academic performance, but rather, it is the GTAs’ professional development in inclusive teaching that may ultimately foster equitable student achievement in science laboratory courses (Wheeler et al., 2017; Lee and Ing, 2020). GTAs play an important role in guiding first-year students and nonmajors through their prerequisite course work. For undergraduate biology students, STEM GTAs in laboratory courses may be the most accessible instructors in their collegiate journey. Therefore, to foster an inclusive and equitable undergraduate student experience in STEM fields, we must consider how to support the pedagogical training of STEM GTAs teaching laboratory courses. Instructor Talk could be a key part of this professional development.

Several researchers have investigated pedagogical training for STEM graduate students (Gormally et al., 2016; Patrick et al., 2016; Lee, 2019). One may question how GTA pedagogical training corresponds to changes in their instruction. A possible outcome to measure may be graduate students’ use of noncontent language in their laboratory courses; however, the nature of GTA Instructor Talk has yet to be characterized. Because GTAs are in a different phase of their careers than their faculty counterparts, they likely navigate classroom teaching differently than faculty. Further, STEM GTAs primarily teach laboratory courses and not lectures, so there is a confounding variable for what is unique to labs or to their career phase. In the present study, we hypothesize that several characteristics of GTAs may influence the nature of their Instructor Talk: GTAs’ near-peer relationship with students, dual identity as graduate students and instructors, and prior teaching experience and pedagogical training.

First, we hypothesize that GTAs’ near-peer relationship with undergraduate students may contribute to the language they use while teaching. Given that undergraduates who persist in STEM majors cite GTAs as more approachable than professors for academic support (Seymour and Hewitt, 1997; Seymour and Hunter, 2019), GTAs may be an exemplar for investigating instructor immediacy. Further, laboratory sections tend to have smaller class sizes than lectures, so GTAs may foster instructor immediacy by offering individualized attention to students. Additionally, according to a survey by the National Postsecondary Student Aid Study, the average age of graduate students in the biological and biomedical sciences in 2007 was among the lowest of various fields of study in the United States—28.5 years (Bell, 2009). Perhaps due to GTAs’ relatively younger age, undergraduate students may perceive GTAs as more relatable, understanding, and able to personalize teaching when compared with faculty instructors in biology (Kendall and Schussler, 2012, 2013). Along with these factors, we predict that GTAs use different kinds of noncontent language when teaching to foster their near-peer relationship with undergraduates. The nature of Instructor Talk used by GTAs could be one explanatory mechanism for student perceptions of GTAs compared with faculty.

Second, despite the benefits GTAs may have with their near-peer relationship and approachability for undergraduates, GTAs generally have less teaching experience relative to faculty. Park and Ramos (2002) described GTAs’ lack of ownership in their job and dual identity as both students and instructors. Further, undergraduates may perceive graduate students to be more hesitant, nervous, and uncertain than their faculty counterparts in biology (Kendall and Schussler, 2012). We hypothesize that this dual identity in relation to GTAs’ teaching roles may impact their Instructor Talk. For example, GTAs may sense a need to exert control and assume poor behaviors of undergraduate students in anticipation of student resistance to lab activities that they did not design or even teach before. These factors suggest that GTAs’ Instructor Talk in lab classrooms would differ from faculty in lecture classrooms.

Third, although GTAs tend to have less teaching experience, they likely encounter a range of professional development opportunities in scientific teaching. Pedagogical training is increasingly available, though widely variable, across institutions (Schussler et al., 2015). Such training can decrease the teaching anxiety felt by GTAs (Chen Musgrove and Schussler, 2022). Further, pedagogical training and prior teaching experience increase the self-efficacy of GTAs and new instructors, both in STEM (DeChenne et al., 2015; Connolly et al., 2018) and across academic departments (Prieto and Altmaier, 1994). Self-efficacy, or the belief in one’s own abilities and capacity to deal with various situations, may impact how GTAs communicate with students. Previous research on the outcomes of pedagogical training for early-stage faculty demonstrated shifts in instructors’ attitudes about students when the training addressed inclusion in STEM classrooms and explored barriers to learning, like stereotype threat (Frey et al., 2020; O’Leary et al., 2020). Therefore, we hypothesize that prior teaching experience and pedagogical training could correspond to variation in GTAs’ Instructor Talk. However, the influence of prior teaching experience and pedagogical training on GTAs’ language used in laboratory classrooms has yet to be studied.

Finally, providing opportunities to reflect on one’s own Instructor Talk is a potential method for encouraging instructors to be metacognitive about the language they use in the classroom and how it might impact students. Metacognition not only encourages our students to learn but could also be a strategy to improve instructors’ teaching (Tanner, 2012). In response to calls for expanding the professional development provided to GTAs (Rushin et al., 1997; Park and Ramos, 2002; Kendall et al., 2013; DeChenne et al., 2015; Chen Musgrove and Schussler, 2022), we anticipate that metacognitive reflection about noncontent instructor language would be an effective intervention for GTA pedagogical training. However, the extent to which GTAs use noncontent language in laboratory courses and their metacognitive reflections on such language have yet to be determined. Findings from a systematic investigation of GTA Instructor Talk could offer a new approach for professional development to better align the language used by GTAs with goals for student learning and inclusion.

Given the importance of GTAs and laboratory courses in undergraduate students’ biology education, we investigated the Instructor Talk used by GTAs in this context to address a gap in the research literature. These investigations were guided by the following research questions:

1. To what extent is Instructor Talk present in biology laboratory classrooms taught by GTAs?

2. To what extent can the Instructor Talk used by GTAs be characterized using the existing Instructor Talk frameworks?

3. How, if at all, does the Instructor Talk used by GTAs teaching in laboratory classrooms differ quantitatively and qualitatively from that previously observed among faculty teaching in lecture classrooms?

4. To what extent does the quantity and type of Instructor Talk in laboratory classrooms taught by GTAs vary depending on whether the GTA has gone through professional development in teaching or has prior teaching experience?

METHODS

To address these research questions, GTAs in a master’s program at a large urban university audio-recorded their laboratory courses across the biology department. Laboratory courses in the study included those in general biology for majors and nonmajors, cell and molecular biology, ecology and evolution, and physiology. The pedagogy of these laboratory courses varied from traditional cookbook labs to inquiry-based labs; no course-based undergraduate research experiences (CUREs) are represented.

Institutional Research Board

This study was determined to be exempt from oversight by the Institutional Review Board at San Francisco State University under project number E18-322 and represents a master’s-level graduate thesis.

Recruitment

A small pilot study was conducted during an initial semester and expanded to a larger study conducted the following semester. For the pilot study, we contacted six GTAs and asked them to audio-record their biology laboratory classrooms. The following semester, we identified 38 GTAs teaching at least one undergraduate laboratory course within the biology department, based on the course schedule and communications with laboratory coordinators. We successfully contacted 37 GTAs and invited them to participate by audio-recording their laboratory courses. A $50 gift card was offered as compensation for completing the required class recordings for the study. Overall, 41 unique GTAs were invited to participate in this research, including both the pilot study (six GTAs invited, five participated) and the main study (37 GTAs invited, two of whom were also in the pilot).

Data Collection

GTAs who agreed to audio-record their courses were issued a standard recorder to be carried in their lab coat pockets and exchanged monthly to allow for frequent data backups. The number of each audio recorder was saved in a password-protected Excel file along with the name of the GTA to ensure that audio files were connected to a pseudonym for each participant. Only the primary researcher, K.A.G., had access to this encrypted Excel file. At the end of the semester, each GTA was asked to complete an optional, short demographic survey. This survey asked GTAs about their pedagogical training, prior teaching experience, gender, race/ethnicity, and whether they were first-generation college-going students. GTAs were excluded from analyses if fewer than 50% of their class sessions were recorded.

Data Analyses

To investigate whether Instructor Talk is present in biology laboratory courses taught by GTAs and to what extent GTAs’ noncontent language can be characterized by existing frameworks, audio recordings were transcribed and instances of Instructor Talk were coded using the methods described in the following sections.

Transcribing and Identifying Instances of Instructor Talk

Audio recordings were only heard by the primary researcher, K.A.G., as they were transcribed. Transcriptions were only associated with pseudonyms. To train the primary researcher, the senior researcher who co-developed the Instructor Talk frameworks with both Seidel et al. (2015) and Harrison et al. (2019), K.D.T., reviewed transcripts with K.A.G. to identify instances of Instructor Talk. Subsequently, K.A.G. transcribed all possible instances of Instructor Talk, ignoring any student or instructor language about biology content. An instance of Instructor Talk is defined as a coherent set of language used by an instructor that is not directly related to the concepts under study but instead focuses on creating the learning environment (Seidel et al., 2015). The transcriptions of all possible instances of Instructor Talk included an oversampling of instances that were not Instructor Talk, suggesting the primary researcher was generous in identifying instances of Instructor Talk in the transcription process. Wherever possible, Instructor Talk instances were split into pieces that could be assigned to a single framework (Positively Phrased or Negatively Phrased), as well as a single category and subcategory. To compare how Instructor Talk used by GTAs teaching laboratory classrooms differs from previously recorded Instructor Talk from faculty teaching lecture classrooms (Harrison et al., 2019), an instance of Instructor Talk was established if the quote could be coded by a single category and subcategory, as defined in previous studies (Seidel et al., 2015; Harrison et al., 2019).

Validating a Sampling Strategy

To collect Instructor Talk among GTAs teaching laboratory courses, we validated a sampling strategy that was previously established by Harrison et al. (2019) for use in lecture classrooms taught by faculty. This sampling strategy was grounded in the observation that the first 15 minutes of a class session will have a representative or enriched sample of Instructor Talk when compared with an entire class session. The samples were even representative of category and subcategory levels of Instructor Talk used for the whole course (Harrison et al., 2019). To assess the validity of this sampling strategy in lab classrooms, class sessions recorded in the initial pilot study were transcribed in their entirety. These class sessions were then analyzed to confirm that this sampling strategy would yield representative or enriched samples of Instructor Talk in laboratory courses. We compared the prevalence of Instructor Talk in the first 15 minutes of a class session with the prevalence of Instructor Talk across an entire class session. Further, to confirm that the sampling method accurately represented instances of Instructor Talk across categories of the frameworks, the entire lab class recordings for the five pilot GTAs were divided into three parts, and categorical instances were quantified in each respective part, which additionally demonstrated that Instructor Talk was indeed concentrated in the beginning of class sessions (see Supplemental Figure A). After confirming the validity of this sampling strategy with laboratory courses, we shifted to transcribing only the first 15 minutes of class session over dozens of courses recorded in the second semester of the study. To compare GTA Instructor Talk in laboratory courses with previously recorded instances of Instructor Talk in lecture courses taught by faculty in Harrison et al. (2019), we used the same sampling method as the previous study. Two 15-minute samples per instructor were transcribed—one from early in the semester and one from the middle of the semester—and then combined into a single 30-minute sample per instructor. All GTAs were instructed to record all class sessions throughout the semester without knowledge of which recordings would be analyzed.

Structure of Instructor Talk Frameworks and Coding Instances of GTA Instructor Talk

To characterize the types and prevalence of GTAs’ Instructor Talk in biology laboratory courses, K.A.G. and K.D.T. coded transcripts using previously developed Instructor Talk frameworks (Table 1). Seidel et al. (2015) and Harrison et al. (2019) developed the Positively Phrased and Negatively Phrased Instructor Talk frameworks from transcribed recordings of faculty teaching lecture courses in undergraduate biology. They analyzed instances of Instructor Talk using a grounded theory approach. Each resulting framework consists of five overarching categories and 15–17 subcategories (see Table 1). Notably, the categories and subcategories of the Negatively Phrased Instructor Talk framework mirror those of the Positively Phrased framework. The five categories of the Positively Phrased Instructor Talk framework are 1) Building the Instructor/Student Relationship, 2) Establishing Classroom Culture, 3) Explaining Pedagogical Choices, 4) Sharing Personal Experiences, and 5) Unmasking Science. The mirroring five categories of the Negatively Phrased Instructor Talk framework are 1) Dismantling the Instructor/Student Relationship, 2) Disestablishing Classroom Culture, 3) Compromising Pedagogical Choices, 4) Sharing Personal Judgment, and 5) Masking Science.

In the present study, each transcribed GTA quote was coded as Instructor Talk or Not Instructor Talk. From the resulting Instructor Talk instances, K.A.G. and K.D.T. independently 1) coded whether the Instructor Talk instance was Positively Phrased or Negatively Phrased, 2) assigned the instance to one of the five categories in the corresponding framework, and 3) categorized the instance into one of the remaining subcategories. Instances of Instructor Talk were assumed to be Positively Phrased wherever possible. Negatively Phrased categories were reserved for Instructor Talk that represented “language that may discourage students or distract from the learning process” (Harrison et al., 2019, p. 2). Each researcher then assigned a category and subcategory, splitting instances into multiple categories or subcategories if necessary, to assign a single code to each instance. The categorization and subcategorization processes were supported using details from the published Instructor Talk framework rubrics (see supplemental material for Harrison et al., 2019) and prior examples of Instructor Talk. Both researchers then met to discuss their assignments and coded each instance to consensus. If an Instructor Talk instance did not fit any category or subcategory, a new category or subcategory would be created.

Post Hoc Quantification and Statistical Analyses

After harvesting and coding instances of Instructor Talk, we quantified and compared the number of different types of Instructor Talk. For each instructor, we quantified the total number of Instructor Talk instances that were coded as Positively versus Negatively Phrased and in each category and subcategory per 30-minute sample. We then calculated the average rate (mean ± SEM) of Instructor Talk used by different groups of instructors, such as GTAs versus faculty, instructors with or without formal pedagogical training, or instructors with or without prior teaching experience. We also compared the average rate (mean ± SEM) of Instructor Talk during the early-semester sample with the average rate of Instructor Talk during the midsemester sample. We tested for normality of the data with the Shapiro-Wilk test and for equal variance with Levene’s test in R (R Core Team, 2019). Because unequal variance was to be expected and normality of the data is not essential for Welch’s t test (Delacre et al., 2017), we performed Welch’s unequal variance t tests. Statistical analyses were conducted in R (R Core Team, 2019). To minimize type I error from multiple statistical comparisons in our category-level analyses, Bonferroni corrections were used for the five category-level comparisons of each Instructor Talk framework (significance level was set at 0.01).

Disaggregation of Instructor Talk Data by GTA Teaching Experience and Pedagogical Training

For the purposes of investigating the impact of prior teaching experience and pedagogical training, we categorized GTAs based on their stated prior teaching experience and pedagogical training on the optional demographic survey. GTAs were disaggregated by teaching experience based on their answer to the question, “Do you have any previous experience teaching prior to this semester?” on the demographic survey. GTAs who answered “yes” were listed as having prior teaching experience; GTAs who answered “no” were listed as having no prior teaching experience. GTAs were also categorized by pedagogical training based their answer to the question, “Do you have any formal training in teaching methods (e.g., courses, credentials, workshops)? Please describe.” GTAs who described having 5 or more hours of formal pedagogical training were grouped as having pedagogical training. GTAs who described having fewer than 5 hours of training in effective teaching methods before the start of the semester were grouped as having minimal or no pedagogical training. GTAs were then disaggregated into these three groups:

1. BOTH prior teaching experience AND prior pedagogical training,

2. teaching experience BUT minimal or no pedagogical training, or

3. NO teaching experience AND minimal or no pedagogical training.

None of the GTAs in the present study fit the classification of “NO teaching experience BUT pedagogical training.” Given that the assumptions of normality were not met with the Shapiro-Wilk test, statistical differences across groups were tested with the Kruskal-Wallis one-way analysis of variance (ANOVA) test. Dunn’s post hoc test (1964) offered pairwise statistical comparisons (significance level set at 0.05). Both Kruskal-Wallis and Dunn’s post hoc test were performed in R (R Core Team, 2019).

Identification of Small-Group Instructor Talk

Instructor Talk is defined as “as any language used by an instructor that is not directly related to the concepts under study but instead focuses on creating the learning environment.” Previously, this was restricted to instances in which the instructor addressed the entire class (Seidel et al., 2015; Harrison et al., 2019). However, we hypothesized that the structure of laboratory courses may provide opportunities to detect Instructor Talk used with small groups of students. We therefore anticipated a new type of Instructor Talk, Small-Group Instructor Talk, which we predicted would include any instance of Instructor Talk addressing a small group of students, rather than the entire class. Instances of Instructor Talk were considered Small-Group Instructor Talk when the context of the language used or the tone and volume of the instructor’s voice made it clear that the instructor was addressing a small group of students or a single student one-on-one during class, rather than the entire class. When there was uncertainty, instances were considered to be addressing the entire class. While Small-Group Instructor Talk potentially could have been present in previous studies, it would not have been captured by the recording methods used. Specifically, the recorder for the lecture courses was kept at the front of the room (Seidel et al., 2015; Harrison et al., 2019), while GTAs were asked to keep recorders for the present study in their pockets. We expected this shift in recording methodology, along with the structure of laboratory courses, might enable the identification of Small-Group Instructor Talk.

GTA Reflection on Instructor Talk Findings

To further engage the participating GTAs in this study, we chose to invite them to discuss the results of this study and reflect on their measured Instructor Talk. Interested and willing GTA collaborators were invited to participate in synchronous meetings and asynchronous reflections on their own Instructor Talk. During the synchronous meetings, GTAs worked in teams to code example instances of Instructor Talk into the five Positively Phrased categories and were asked to reflect on this short experience of qualitative coding. GTAs then received files with their own Instructor Talk samples and were given a few minutes to briefly look through them. The GTAs were then given the opportunity to discuss privately in small groups what their immediate impressions of their Instructor Talk samples were.

After the synchronous meeting, GTAs who were interested in continuing work on this project completed written reflections about their experiences with the project and their own Instructor Talk samples. They were asked to share their response to the following prompt:

To ensure trustworthiness and independent identification of emergent themes, two coders (K.A.G. and K.D.T.) independently read all GTA reflections and developed their own thematic categories. The two coders came to consensus agreement on the resulting categories, and all language of GTA reflections was discussed and coded into at least one emergent theme. The relative brevity of the reflections, ranging from approximately 30 to 600 words, and the low number of instances of each emergent theme precluded statistical analysis. We will show several examples of themes detected in these reflections.

RESULTS

Participant Population

This study analyzes the data collected from GTAs teaching laboratory courses and compares them with previously published data on faculty teaching lecture courses (Harrison et al., 2019). A summary of the participant population in this study and the participant population of the faculty we used as a comparison is shown in Table 2. All six GTAs who were invited for the pilot study and 20 out of 37 GTAs who were invited for the full study agreed to participate. Five of the six GTAs from the pilot study and 19 of the 37 GTAs from the full study successfully recorded more than 50% of their class sessions and were included in analysis. Two GTAs participated in both studies. This resulted in 22 unique GTAs who audio-recorded 24 unique courses across both semesters.

Validation of a Sampling Method in the Pilot Study

We aspired to gauge the quantity and character of Instructor Talk across a large number of courses. To that end, we validated a previously established sampling strategy in the novel context of laboratory classrooms taught by GTAs. Based on the pilot study of five GTAs’ entire lab class recordings that were then divided into three parts, the sampling method was validated to produce representative or enriched profiles of Instructor Talk instances across categories (see Supplemental Figure A). Additionally, we replicated the sampling strategy used by Harrison et al. (2019), which examined only the first 15 minutes of a class session with the goal of providing a representative or enriched sample of Instructor Talk. We confirmed that this was the case in the pilot study by comparing the actual percentage of Instructor Talk that was found in the first 15 minutes with the expected percentage if Instructor Talk were evenly distributed across the class session. The expected percentage of Instructor Talk that would be found in the first 15 minutes of a class session, if Instructor Talk instances were evenly distributed, was calculated by dividing 15 minutes by the length of the class session. For example, if a class session was 60 minutes long, we would expect 25% of the Instructor Talk instances to occur during the first 15 minutes if they were evenly distributed. We compared this with the actual percentage of Instructor Talk instances that occurred during the first 15 minutes using two class sessions for each of the five courses recorded during the initial semester.

As shown in Figure 1, the actual percentage of Instructor Talk that occurred in the first 15 minutes was greater than the expected percentage of Instructor Talk for all five courses (61.0% actual vs. 19.6% expected; 50.7% actual vs. 20.3% expected; 44.8% actual vs. 22.6% expected; 38.2% actual vs. 13.3% expected; 36.9% actual vs. 14.1% expected), suggesting this sampling method would provide representative or enriched samples of Instructor Talk. Therefore, we used this sampling method for the remainder of our analyses. We transcribed two 15-minute samples per instructor—one from early in the semester and one from the middle of the semester. We then combined these early- and midcourse samples into a single 30-minute sample per instructor.

Determining Whether Instructor Talk Is Present in Laboratory Classrooms Taught by GTAs

Instructor Talk was found in all 24 courses studied, though the rate of use was highly variable among instructors. Figure 2A depicts the total number of Instructor Talk instances in each instructor’s combined 30-minute sample. Figure 2B and C shows the total number of Instructor Talk instances broken down into the 15-minute early- and midcourse samples, respectively, for each instructor. The average rate of Instructor Talk observed per 30 minutes across all 24 courses (mean ± SEM) was 14.0 ± 1.9 instances, and the range was between two and 44 instances per 30 minutes. Both Positively Phrased and Negatively Phrased Instructor Talk were found in laboratory classrooms taught by GTAs, with an average rate of use of Positively Phrased Instructor Talk of 11.8 ± 1.6 instances per 30 minutes and an average rate of use of Negatively Phrased Instructor Talk of 2.3 ± 0.6 instances per 30 minutes. Tables 3 and 4 provide examples of Positively Phrased and Negatively Phrased Instructor Talk instances, respectively, as used by GTAs teaching laboratory courses.

Comparing the Average Quantity of Instructor Talk Used by GTAs Teaching Laboratory Courses with Faculty Teaching Lecture Courses

The average rate of use of Instructor Talk by GTAs teaching laboratory courses (n = 24) was compared with that used by faculty teaching lecture courses (n = 61) in terms of overall Instructor Talk use, Positively Phrased Instructor Talk, and Negatively Phrased Instructor Talk across all categories and subcategories. For some of the categories being compared, the Shapiro-Wilk test showed a nonnormal distribution, and Levene’s test showed unequal variance. For comparing groups of unequal variance with assumptions of normality not met (Delacre et al., 2017), we performed Welch’s unequal variance t tests. Further, to test the robustness of our chosen statistical analyses to the underlying sample distributions, we also performed the nonparametric Mann-Whitney test on each of these comparisons. For all comparisons, we obtained similarly significant p values (unpublished data).

As shown in Figure 3, the average rate of Instructor Talk used by GTAs teaching laboratory courses was significantly higher than by faculty teaching lecture courses (p = 0.03, Welch’s unequal variances t test, significance set at 0.05). Specifically, GTAs used 14 ± 1.9 instances per 30 minutes, and faculty used 9.3 ± 0.9 instances per 30 minutes of instruction. The mean and SEM are shown as a bar graph in Figure 3A, and the distribution of individual data points is shown in a box-and-whisker-plot in Figure 3B. Table 5 lists the overall average rate of Instructor Talk use by GTAs and faculty and the rates of use of Positively and Negatively Phrased Instructor Talk.

As shown in Figure 4, the rate of Positively Phrased Instructor Talk observed by GTAs teaching laboratory courses was compared with faculty teaching lecture courses. The overall rate of use of Positively Phrased Instructor Talk was 12 ± 1.6 instances per 30 minutes for GTAs and 8.5 ± 0.9 instances per 30 minutes for faculty. Figure 4 compares the average rate of use of Instructor Talk by GTAs and faculty for each of the Positively Phrased Instructor Talk categories (Figure 4A) and subcategories (Figure 4B). No significant differences were observed when comparing the rate of Positively Phrased Instructor Talk at the category level between GTAs in laboratory courses and faculty in lecture settings (see Table 6 for p values, Welch’s unequal variances t test with Bonferroni corrections for five categorical comparisons, significance set at 0.01). Two subcategories showed more Instructor Talk use by GTAs: Revealing Secrets to Success within the Building the Instructor/Student Relationship category and Preframing Classroom Activities within the Establishing Class Culture category. GTAs used Instructor Talk that could be categorized as Revealing Secrets to Success at 1.8 ± 0.4 instances per 30 minutes, while faculty used 0.7 ± 0.2 instances per 30 minutes. Furthermore, GTAs used 3.0 ± 0.4 instances of Instructor Talk per 30 minutes that could be categorized as Preframing Classroom Activities, while faculty used 1.3 ± 0.2 instances per 30 minutes. While these two subcategories showed increased use by GTAs, the number of instances observed at the subcategory level precludes comparative statistical analyses. GTAs used more Positively Phrased Instructor Talk, but this was not statistically significant for any of the Positively Phrased Instructor Talk categories (see Figure 4). The Positively Phrased Instructor Talk category-level data are summarized in Table 6 and the subcategory-level data are summarized in Table 7.

As shown in Figure 5, the overall rate of use of Negatively Phrased Instructor Talk was 2.3 ± 0.6 instances per 30 minutes for GTAs and 0.8 ± 0.2 instances per 30 minutes for faculty. One category, Dismantling the Instructor/Student Relationship, showed a statistically significant difference. For this category, GTAs used 1.3 ± 0.4 instances per 30 minutes, while faculty used 0.1 ± 0.03 instances per 30 minutes (p < 0.01, Welch’s unequal variances t test with Bonferroni corrections for five categorical comparisons, significance set at 0.01). Furthermore, this increase at the category level for Dismantling the Instructor/Student Relationship may be caused by an increased use of the subcategory Assuming Poor Behaviors, for which GTAs used 1.0 ± 0.3 instances per 30 minutes; however, instances of this subcategory were not found in the faculty sample. For the remaining four categories of the Negatively Phrased Instructor Talk framework, no significant differences were observed when comparing GTAs in laboratory courses and faculty in lecture settings (see Table 8 for p values, Welch’s unequal variances t test with Bonferroni corrections for five categorical comparisons, significance set at 0.01). Subcategory-level use of Negatively Phrased Instructor Talk is shown in Figure 5B; however, the number of subcategory observations precludes comparative statistical analyses. Figure 5 shows the rate of use of all five Negatively Phrased categories (Figure 5A) and 15 subcategories (Figure 5B). The Negatively Phrased category-level data are summarized in Table 8 and the subcategory data in Table 9.

Comparing Instructor Talk Used by GTAs Based on Their Prior Pedagogical Training and Prior Teaching Experience

As shown in Figure 6, GTAs with prior teaching experience and pedagogical training used more Instructor Talk than GTAs without prior teaching experience and with minimal or no pedagogical training. Figure 6 illustrates the average rate of use of Instructor Talk for total Instructor Talk, Positively Phrased Instructor Talk, Negatively Phrased Instructor Talk, early-course sample, and midcourse sample for 1) GTAs with both teaching experience and prior pedagogical training (n = 11); 2) GTAs with teaching experience with minimal or no prior pedagogical training (n = 8); and 3) GTAs with no teaching experience and minimal or no prior pedagogical training (n = 5). Table 10 summarizes this data along with p values as calculated by Kruskal-Wallis one-way ANOVA.

GTAs with both prior teaching experience and pedagogical training used 19 ± 3.2 instances of Instructor Talk per 30 minutes of instruction; GTAs with prior teaching experience but minimal or no pedagogical training used 12 ± 2.0 instances per 30 minutes; GTAs with neither prior teaching experience nor pedagogical training used 6.6 ± 1.9 instances per 30 minutes. This was found to be statistically significant (Kruskal-Wallis, χ² = 7.1, p = 0.029). Dunn’s post hoc test detected a statistically significant difference between GTAs with neither pedagogical training nor teaching experience and GTAs with both training and experience (p = 0.027, adjusted with Bonferroni method).

While we observed significantly higher levels of Positively Phrased Instructor Talk for instructors with teaching experience and pedagogical training, it was less clear for Negatively Phrased Instructor Talk. GTAs with both prior teaching experience and pedagogical training used 16 ± 2.4 instances of Positively Phrased Instructor Talk and 3.4 ± 1.3 instances of Negatively Phrased Instructor Talk per 30 minutes of instruction. GTAs with prior teaching experience but minimal or no pedagogical training used 11 ± 2.1 instances of Positively Phrased Instructor Talk and 1.1 ± 0.4 instances of Negatively Phrased Instructor Talk per 30 minutes of instruction. GTAs without teaching experience and minimal or no pedagogical training used 5.0 ± 1.7 instances of Positively Phrased Instructor Talk and 1.6 ± 0.7 instances of Negatively Phrased Instructor Talk per 30 minutes of instruction. This difference was statistically significant for Positively Phrased Instructor Talk (Kruskal-Wallis, χ² = 8.2, p = 0.017), but not for Negatively Phrased Instructor Talk (Kruskal-Wallis, χ² = 0.56, p = 0.75). Dunn’s post hoc test detected a statistically significant difference in Positively Phrased Instructor Talk between GTAs with neither pedagogical training nor teaching experience and GTAs with both training and experience (p = 0.013, adjusted with Bonferroni method).

There was also a statistically significant difference in the amount of Instructor Talk used during the early-course and midcourse samples. The early-course sample had 12 ± 2.5 instances per 15 minutes for GTAs with both prior teaching experience and pedagogical training, 5.0 ± 1.0 instances per 15 minutes for GTAs with prior teaching experience but minimal or no pedagogical training, and 3.8 ± 1.6 instances per 15 minutes for GTAs without teaching experience and minimal or no pedagogical training (Kruskal-Wallis, χ² = 7.4, p = 0.024). The midcourse sample had 6.7 ± 1.0 instances per 15 minutes for GTAs with both prior teaching experience and pedagogical training, 6.8 ± 1.5 instances per 15 minutes for GTAs with prior teaching experience but minimal or no pedagogical training, and 2.8 ± 0.5 instances per 15 minutes for GTAs without teaching experience and minimal or no pedagogical training (Kruskal-Wallis, χ² = 6.5, p = 0.040). Interestingly, for the midcourse sample, Dunn’s post hoc test detected a statistically significant difference between GTAs with neither pedagogical training nor teaching experience and GTAs with both training and experience (p = 0.038, adjusted with Bonferroni method), but this comparison was not significant for the early-course sample (p = 0.054, adjusted with Bonferroni method). These results are also summarized in Figure 6 and Table 10.

A Potentially New Type of Instructor Talk: Small-Group Instructor Talk in Pilot Study

Due to the nature of laboratory classrooms, instructors have greater opportunity to walk around the room and work with groups of students, and we hypothesized that there might be a new type of Instructor Talk: Small-Group Instructor Talk. We found that only a small percentage of the total Instructor Talk instances were addressed to a small group of students rather than the entire class. The percentage of total Instructor Talk instances that were addressed to a small group of students for each of the courses in the initial pilot semester were: Xavier (2%), Gabriel (6%), Vanessa (8%), Jon (14%), and Yolanda (22%). As such, we did not continue to investigate Small-Group Instructor Talk in the subsequent semester and in the full study.

Qualitative Analyses of GTA Reflections on Their Own Instructor Talk

After reviewing their own Instructor Talk, the GTA participants reflected on what they learned from looking at their Instructor Talk and how it might impact their teaching in the future. Responses were received from 17 of the 22 GTA instructors (77% participation rate).

Five themes emerged from these reflections, and excerpts from these reflections are included in Table 11.

The five emergent themes and their prevalence among GTAs were: Intending a Future Focus on Instructor Talk (n = 10/17), Acknowledging Negatively Phrased Instructor Talk (n = 8/17), Aspiring to Use Specific Categories of Instructor Talk (n = 3/17), Reporting Little Learned from Instructor Talk (n = 3/17), Differing Instructor Talk across Different Courses (n = 1/17). Up to three examples of GTA reflections related to each of these emergent themes are provided as excerpts in Table 11. Specifically, 10 GTA reflections of the 17 included themes stating that seeing their Instructor Talk sample will affect the language they use in the classroom in the future. Additionally, eight reflections noted that seeing their Negatively Phrased Instructor Talk caused them to reflect on the importance of phrasing and how words are perceived. Furthermore, three reflections named specific categories of Instructor Talk that they wanted to use more of in the future. One GTA’s Instructor Talk was different in two courses the GTA taught, which the GTA attributed to a difference in confidence level when teaching those two courses. Finally, three GTAs expressed that there is little or nothing from this experience that will affect their teaching going forward.

DISCUSSION

This study addresses a gap in the literature about the quantity and nature of Instructor Talk used by GTAs in laboratory courses. It was therefore an exciting find that Instructor Talk is indeed present in this new context. Prior studies have investigated Instructor Talk in lecture courses taught by faculty (Seidel et al., 2015; Harrison et al., 2019; Lane et al., 2021; Ovid et al., 2021) and postdocs (Seah et al., 2021). This study also addresses a gap in the literature regarding the use of Instructor Talk by instructors who have not completed prior pedagogical training or have not had prior teaching experience. Here, we discuss our findings about noncontent language used by GTAs in laboratory courses, comparisons with faculty teaching lecture courses, the impact of GTA pedagogical training and prior teaching experience on Instructor Talk, and the surprising lack of Small-Group Instructor Talk observed in the pilot study.

Noncontent Language Used by GTAs in Laboratory Classes Can Be Characterized Using the Existing Instructor Talk Frameworks, Regardless of GTAs’ Pedagogical Training or Previous Teaching Experience

It was unclear whether graduate students would use noncontent language in their laboratory classes and whether such language could be categorized with the existing Instructor Talk frameworks. To our surprise, GTAs used noncontent language in laboratory classes that reflected the categories of the existing Instructor Talk frameworks. While emerging frameworks have been developed for measuring Instructor Talk that may be unique to the first day of lecture courses (Lane et al., 2021; Meaders et al., 2021) or online teaching during the pandemic (Seah et al., 2021), the present study demonstrates that the Instructor Talk frameworks originally developed for biology lecture courses can be used to categorize Instructor Talk used by GTAs in laboratory courses. Importantly, we observed from our pilot study that Instructor Talk used by GTAs in the first 15 minutes overrepresented the amount of Instructor Talk used throughout the entire class session, indicating that the previously developed sampling method (Harrison et al., 2019) could be used across a large sample of laboratory courses. Further, our findings fill a gap in the literature by investigating the Instructor Talk of biology instructors with little to no pedagogical training. Interestingly, even GTAs with no formal pedagogical training used Instructor Talk that could be characterized with the existing frameworks, suggesting that these frameworks could be used to investigate Instructor Talk of untrained instructors. Additionally, future studies could explore the applicability of Positively Phrased and Negatively Phrased Instructor Talk frameworks in the context of other courses commonly led by GTAs, such as discussion sections, or even different kinds of laboratory courses, such as CUREs.

Increased Negatively Phrased Instructor Talk Use by GTAs Teaching Laboratory Classes May Reflect GTA Dual Identity as both Student and Instructor

GTAs in lab classrooms used more Negatively Phrased Instructor Talk than did faculty in lecture classrooms. This trend was most clear in the category Dismantling the Instructor/Student Relationship and associated subcategory Assuming Poor Behaviors. Forty-two percent of all of the Negatively Phrased Instructor Talk used by GTAs fell in the subcategory Assuming Poor Behaviors, which included assuming that students do not want to be in class, participate in activities, or learn, as well as assuming that students were sleeping in class, not doing their homework, or not studying. In contrast, this subcategory was not observed at all in a prior faculty sample (Harrison et al., 2019). Here we explore preliminary hypotheses for how GTA identity might influence this increased use of Negatively Phrased Instructor Talk.

GTAs’ unique position in the classroom may be a potential driver of Negatively Phrased Instructor Talk. GTAs are historically underpaid and given little influence over the content that they are expected to teach (Park and Ramos, 2002; Muzaka, 2009). Increased use of Negatively Phrased Instructor Talk may reflect GTAs themselves not wanting to lead specific classroom activities that they do not have the agency to change. GTAs have dual identities as both students and instructors, while being expected to effectively teach undergraduates who are often only a few years behind them academically and rarely being given agency in the curriculum they are expected to teach (Park and Ramos, 2002). Further, one might consider how GTAs anticipate and pre-empt the phenomenon of student resistance (Seidel and Tanner, 2013). Given the lack of agency in course design, GTAs may be expressing to undergraduates that they can empathize with them and are “in the know” about student resistance, accept it as inevitable, and therefore take little to no responsibility for student engagement. One wonders if this dual identity and potential reduced agency in their classrooms may result in increased use of Negatively Phrased Instructor Talk that signals that no one wants to be in class, such as:

GTAs’ dual identity as both instructors and students may play another role in the Instructor Talk they use, due to the power dynamic it creates. Because of their status as students themselves, GTAs are more likely to be early in their careers and to be new to teaching. Of the 42% of Negatively Phrased Instructor Talk instances that fell within the subcategory Assuming Poor Behaviors, many asserted the GTA’s authority with a “don’t mess with me” attitude, such as:

Instances such as these show that GTAs may be concerned about being taken advantage of by students because of their status. It is possible that GTAs’ difference in position compared with faculty could also lead some GTAs to respond by overcorrecting and asserting their authority as the instructor by using Negatively Phrased Instructor Talk. Previous research has explored GTAs’ ability to establish power with their students, stating that GTAs may be too inexperienced to be able to effectively communicate their credibility and power (Pytlak and Houser, 2014). Additionally, it has been found that GTAs are more “hesitant, nervous, uncertain, and unsure how to begin teaching” (Kendall and Schussler, 2012, 2013). We posit that GTAs may respond to the ambiguous position of being both instructors and students and their uncertainty of how to manage being instructors by using Negatively Phrased Instructor Talk that assumes poor behaviors from their students.

Such language may be common for instructors new to teaching, not just GTAs. Further research could investigate Instructor Talk used by instructors at varying points in their careers or could investigate correlations between GTA perceptions of teaching, professional identity, and their use of Instructor Talk.

GTAs Teaching Laboratory Courses May Use More Positively Phrased Instructor Talk Than Previously Observed with Faculty Teaching Lecture Courses

Our study found a trend of GTAs using more Positively Phrased Instructor Talk than previous observations found with faculty. We explore here two potential reasons—near-peer status and laboratory course structure—that may contribute to these findings and that could guide future research. Two subcategories showed increased Instructor Talk use by GTAs: Preframing Classroom Activities within the Explaining Pedagogical Choices category and Revealing Secrets to Success within the Building the Instructor/Student Relationship category.

Increased GTA use of Positively Phrased Instructor Talk may relate to GTAs’ near-peer status with their undergraduate students. One place where the increase in Instructor Talk use was particularly clear was the subcategory Revealing Secrets to Success and corresponding category Building the Instructor/Student Relationship. Through this type of Instructor Talk, GTAs shared study strategies or other tips for succeeding in learning or in life. GTAs used more than double the number of Instructor Talk instances than faculty in this particular subcategory; however, with the small number of total instances present at the subcategory level, this was not statistically significant. We predict that some of this apparent increase in Positively Phrased Instructor Talk may be related to the fact that GTAs are near peers to their undergraduate students and may better identify with the struggles their students are encountering than faculty instructors (Muzaka, 2009). Undergraduate students perceive their GTA instructors as more relatable, understanding, and engaging than their faculty counterparts (Kendall and Schussler, 2012). It may be the case that GTAs maintain a near-peer status through their Instructor Talk by promoting instructor immediacy. Given undergraduate students’ heightened sense of relatability to GTAs, perhaps an affinity bias leads undergraduates to be more receptive to supportive messaging from GTAs. Future research could explore the interplay between Instructor Talk and instructor near-peer status to expand our knowledge of how an instructor’s identity affects the way the instructor uses language in the class and students’ receptivity to such language.

Another potential factor in our observed increased use of Positively Phrased Instructor talk is the laboratory course structure. GTAs also use significantly more Instructor Talk in the subcategory Preframing Classroom Activities and associated category Establishing Class Culture. This subcategory includes preparing students for in-class activities and sharing how much time activities will take or the activity structure. Given the hands-on nature of laboratory courses, this particular increase in Instructor Talk may have been due to the course type—laboratory versus lecture—rather than the instructor type. However, this is unclear, as some lecture courses may include substantial active learning, which would also be framed by Instructor Talk. Further, some laboratory courses might simply involve following lab manuals without instructor input. Without more data on the nature of the courses we studied, we are unable to resolve whether this increase was due to instructor type or course type. Future investigations might compare the Instructor Talk of GTAs in discussion sections compared with laboratory courses.

GTA Teaching Experience and Pedagogical Training Were Associated with More Instructor Talk

Among GTAs with teaching experience and prior pedagogical training, there was a higher rate of both Positively Phrased and overall Instructor Talk use when compared with GTAs with no teaching experience and minimal or no prior pedagogical training. We hypothesize that prior teaching experience and pedagogical training may both work to increase GTAs’ self-efficacy, in turn creating the increased use of Positively Phrased Instructor Talk that was seen. GTAs with more experience in the classroom may be more likely to feel confident in their abilities and feel a greater sense of self-efficacy (DeChenne et al., 2015). This may enable GTAs to focus not only on what they are teaching, but also how they are teaching. As the present study did not measure self-efficacy, future studies could consider correlating GTA Instructor Talk with validated measures in graduate student teaching self-efficacy (Boman, 2008, 2013).

Further, previous research has shown that pedagogical training can increase self-efficacy for GTAs (Prieto and Altmaier, 1994; DeChenne et al., 2015). Increased self-efficacy due to participating in pedagogical training could further contribute to the increased Instructor Talk that is seen. In addition, instructors of STEM pedagogical training may model using Positively Phrased Instructor Talk that GTAs may then mimic in their own classrooms. Additionally, a previous study on pedagogical training of active learning demonstrated the value of GTAs practicing the new skill themselves (Patrick et al., 2016), suggesting that GTAs might practice Positively Phrased Instructor Talk with one another to gain self-efficacy and practice with this language in their own courses.

Both prior teaching experience and pedagogical training were associated with increased amounts of Positively Phrased and overall Instructor Talk for GTAs. Future research may be able to disentangle the effects of prior teaching experience and pedagogical training that we were unable to address in the current study. Given previous research on early-stage faculty training about phenomena such as stereotype threat in STEM (Frey et al., 2020; O’Leary et al., 2020), one wonders how pedagogical training for STEM GTAs that focuses on inclusion could employ Instructor Talk to mitigate stereotype threat and foster equitable student outcomes. Future studies might consider student outcomes in biology laboratory courses using validated psychometric measures (e.g., Wineinger et al., 2021) correlated with GTAs’ Instructor Talk. Our findings of increased Instructor Talk among GTAs with pedagogical training provided yet another call for systematic integration of training in effective and inclusive science teaching for all scientific trainees.

Reflecting on Their Instructor Talk May Provide Instructors an Opportunity for Metacognition

After discussing Instructor Talk, reviewing their own Instructor Talk, and reflecting individually in writing on their own, 82% of GTAs asserted that they had learned something about their teaching and that they had actionable items for improving their future Instructor Talk. There have been calls to provide more pedagogical training to GTAs and improve pedagogical training already available to GTAs for evidence-based teaching practices (Rushin et al., 1997; Park and Ramos, 2002; Kendall et al., 2013; DeChenne et al., 2015). Our findings provide another avenue for GTA pedagogical development, namely, an opportunity to be metacognitive about the language they use in the classroom (Tanner, 2012). While three GTAs did not mention learning anything that would impact their future teaching, 14 GTAs noted specific things they learned from reviewing their Instructor Talk and mentioned actionable items that they would be taking with them into their future teaching. Additionally, the Instructor Talk categories provided GTAs with an analytical framework with which they could identify specific areas for improvement or changing the language they used in their classrooms. Three GTAs noted specific categories in which they intended to use more Instructor Talk in the future to better align their classroom environments with their intentions.

GTAs noted remembering some of their Positively Phrased Instructor Talk, stating that those instances were said intentionally. Many GTAs were surprised by their Negatively Phrased Instructor Talk, sharing that their Negatively Phrased Instructor Talk was intended positively, often either as a joke or sarcasm, and that they could now see how their phrasing could lead to misinterpretations of their intentions. Humor used by instructors may be beneficial in creating a positive classroom environment and building the instructor–student relationship; however, humor that students perceive as offensive may disestablish the instructor–student relationship and decrease student sense of belonging (Cooper et al., 2018). Therefore, professional development that encourages GTAs to be metacognitive about the language they use and how they employ humor may be beneficial. Furthermore, such reflection may support instructors in making the small changes to their phrasing needed to use Positively Phrased Instructor Talk in their jokes, sarcasm, and casual language. Future studies might consider how undergraduate students perceive GTAs’ use of what researchers have categorized as Negatively Phrased Instructor Talk.

Surprisingly, Minimal Small-Group Instructor Talk Was Observed among GTAs Teaching in the Laboratory Course Setting

Due to the nature of laboratory courses, in which students often work in small groups on laboratory activities, we expected to observe Small-Group Instructor Talk, i.e. any Instructor Talk that addresses a small group of students rather than the entire class. However, we found very few instances of Small-Group Instructor Talk in the pilot study, which monitored entire class sessions for five GTAs. Specifically, we transcribed and coded two class sessions—one on the first day and another midsemester—for each GTA in the pilot study. It is possible that there are more instances of Small-Group Instructor Talk to be discovered. Importantly, our sampling method for the full study focused on the first 15 minutes of the class session, at which point GTAs may not have been engaged in small-group conversations. As such, future studies investigating Small-Group Instructor Talk in laboratory courses might consider validating a sampling method for the middle of the class session. Further, it may be the case that GTAs may not be prepared to do small-group instruction, thus limiting the frequency of GTAs’ small-group engagement. Notably, some GTAs were having conversations with small groups of students, but this language did not fall within the Instructor Talk framework. It is also possible that language that falls on the edge of content may not be codable based on our criteria for Instructor Talk.

This type of language has been studied using the Classroom Discourse Observation Protocol, which focuses on language that directly relates to content, but involves the instructor engaging students “in the construction, justification, and evaluation of knowledge as opposed to simply providing factual knowledge” (Kranzfelder et al., 2019, p. 2). One might predict that Small-Group Instructor Talk could be captured in courses structured with active-learning and group work if instructors carry a recorder with them (e.g., in their pockets). However, this type of discourse may be commonly used with small groups of students who may be asking content- or procedure-specific questions, and therefore would not be categorized as Instructor Talk. While it is possible that communication and guidance with small groups of students primarily falls outside the Instructor Talk framework, it is also possible that Small-Group Instructor Talk occurs infrequently. This could imply that the one-on-one attention we predict exists in laboratory settings is not as prevalent as assumed. Alternatively, this lack of Small-Group Instructor Talk could be unique to GTAs who may be more hesitant in their teaching ability and therefore less likely to feel comfortable talking to students in small groups. Although Small-Group Instructor Talk was not widely prevalent in our pilot study, future investigations might consider this phenomenon in contexts where small-group activities are common, such as in CUREs or biology courses with active-learning components.

Limitations and Future Directions

As alluded to earlier, there was a fundamental conflation of variables inherent in our study: course type and instructor type. Our study investigated Instructor Talk used by GTAs teaching laboratory courses and compared it with Instructor Talk of faculty teaching lecture courses. A research design that could separate these variables would compare the current findings to with results from faculty teaching laboratory courses and/or GTAs teaching lecture courses. However, graduate students typically only teach discussions or laboratories and rarely teach lecture courses. Furthermore, in a 4-year research university setting, it is uncommon to find faculty teaching laboratory courses. While this statistic has not been recently examined, national surveys of more than 70 top institutions have found that between 71 and 93% of institutions’ laboratory courses are primarily instructed by GTAs (Sundberg and Armstrong, 1993; Sundberg et al., 2005). Future studies are needed in other settings, such as community colleges, liberal arts universities, and undergraduate-only universities, in which Instructor Talk could be studied with faculty members teaching laboratory courses to address this limitation of the current study.

While we have discovered that Instructor Talk is indeed used by GTAs in laboratory courses, our findings may not generalize to all GTAs in all contexts. This study was restricted to the specific context of GTAs teaching laboratory classes at a large, urban 4-year university. A logical next step would be to investigate Instructor Talk in novel contexts such as other institution types with varied GTA populations and training programs. Furthermore, the GTAs involved were those who agreed to audio-record their courses, and it is possible that Instructor Talk of nonparticipating GTAs may be different from that studied here or even not present at all. Finally, this study investigated Instructor Talk from an instructor-centric point of view, and future studies are needed to investigate the impacts of Instructor Talk on undergraduate students by considering both their academic and affective outcomes, as well as student memories and perceptions of GTA Instructor Talk.

CONCLUSION

We set out to address a gap in the literature—knowledge about the noncontent language used by GTAs teaching laboratory courses—that could have impacts on equity and inclusion in biology education and on the student experience. Our findings here expand upon previous Instructor Talk research. We found that Instructor Talk is indeed found in the novel context of laboratory courses taught by GTAs and can be characterized using the existing frameworks for Instructor Talk. Further, this study fills a gap in the literature by exploring Instructor Talk for biology instructors with little to no pedagogical training or experience. We found that GTAs with prior teaching experience and pedagogical training used more Instructor Talk—specifically more Positively Phrased Instructor Talk—than GTAs without teaching experience and pedagogical training, reinforcing the importance of providing pedagogical training for GTAs. Additionally, we found that GTAs teaching laboratory courses use more Instructor Talk—both Positively and Negatively Phrased—than faculty teaching lecture courses. Future research could explore important aspects of GTA identity, including near-peer status and self-efficacy, and how they relate to Instructor Talk. Overall, our results provide evidence that Instructor Talk may vary at different science instructor career stages. The present study impels future work to investigate the prevalence of noncontent language (or lack thereof) for instructors with minimal pedagogical training and the impacts of Instructor Talk on in/equitable student outcomes in STEM.

ACKNOWLEDGMENTS

REFERENCES