1. Introduction
My scholarly interests have always been on L2 classroom interaction and specifically, the links between teaching and learning as they are accomplished in interaction. My early work drew on a sociocultural perspective of language and learning based on theoretical insights and research from the fields of linguistic anthropology (Hymes, Reference Hymes, Pride and Holmes1972; Ochs, Reference Ochs1988; Ochs & Schieffelin, Reference Ochs, Schieffelin, Duff and Hornberger2007) and cultural psychology (Vygotsky, Reference Vygotsky1978; Tomasello, Reference Tomasello2003). A major premise of a sociocultural perspective is that the process of language development is embedded in the process of becoming socialized into competent participation in the social activities of one's social groups by more knowledgeable members. Individuals are socialized through language to use language.
An implication of this perspective for L2 classrooms is that interactions between teachers and students ‘are consequential in the creation of effectual learning environments and ultimately in the shaping of individual learners’ language development’ (Hall & Walsh, Reference Hall and Walsh2002, p. 186). Drawing on this perspective, in my initial studies of teacher-student interactions in two Spanish foreign language classrooms (Hall, Reference Hall1995, Reference Hall1997) I found, similar to studies on first language classrooms (Mehan, Reference Mehan1979; Cazden, Reference Cazden1988; Wells, Reference Wells1993), that the IRF (Initiation, Response, Feedback) was a ubiquitous pattern of interaction. The IRF is a teacher-led three-action sequence. It opens with an initiation typically in the form of a teacher question or directive to the cohort of students that serves to direct their attention to the focus of instruction (I). The initiation projects the second action, a student response (R), whose function is to display the students’ understanding of what they are expected to do and know. The response projects the third action (F) which serves to evaluate and/or provide feedback on the adequacy of the student response.
Interpretations of the value of the IRF to L2 student learning have varied. Some have argued that its use limits students' opportunities for participation, and more generally, facilitates teacher control rather than student learning (e.g., Waring, Reference Waring2008). Others have argued that the IRF is instrumental to teaching and learning (e.g., Seedhouse, Reference Seedhouse2004; Lee, Reference Lee2007) as each iteration of the IRF responds to the ‘immediate local and pedagogic contingencies and goals of the unfolding lesson’ (Gardner, Reference Gardner2019, p. 214; see Hall, Reference Hall, Haneda and Nassiji2019a, for a more extensive review of the IRF). Nonetheless, despite long held assumptions about the links between L2 teaching and L2 language development underpinning studies on the IRF, until recently there has been little empirical evidence linking the linguistic designs of teaching actions in the IRF and other instructional activities to L2 learners’ developing linguistic repertoires. This is changing, however, as research from fields such as child language development (e.g., Tomasello, Reference Tomasello2003), psycholinguistics (e.g., MacWhinney, Reference MacWhinney2006, Reference MacWhinney, Cadierno and Eskildsen2015), neurolinguistics (e.g., Lee, Mikesell, Joaquin, Mates, & Schumann, Reference Lee, Mikesell, Joaquin, Mates and Schumann2009) and different branches of functional and cognitive linguistics (e.g., Bybee, Reference Bybee2006; Goldberg, Reference Goldberg2006; Larsen-Freeman & Cameron, Reference Larsen-Freeman and Cameron2008; Ellis, Reference Ellis and Robinson2012) has provided robust findings on the usage-based foundations of language knowledge, and specifically on the consequential role of the input to which L2 learners are exposed to their language development.
A great deal of research on L2 input has provided useful linguistic evidence for the development of generalizations about and hypotheses on the usage-based nature of L2 knowledge. However, despite the fact that classrooms are one of the most ubiquitous sites of L2 learning, and despite the myriad studies on the IRF, we still know very little about the linguistic quality of naturally-occurring classroom interactions and their consequential role in shaping learners’ linguistic repertoires (Ellis, Reference Ellis2016). These data are needed in order to understand how differences in L2 classroom interactional activities affect L2 learner language.
This is the focus of the set of research tasks I lay out in this paper. By making clear the interrelationships between teaching and learning, findings from the studies will offer new insights into L2 pedagogy and the key role that L2 teachers play in designing the actions and action projects of their learning contexts. Not only are teacher actions significant to the meaning-making resources that are available to learners, but as importantly, they give shape to the ways in which L2 learners orient to and subsequently use the resources in the design of their own worlds (Hall, Reference Hall2019b). In what follows, after a brief overview of usage-based research on language and a summary of the methodological contributions of CA and IL to such research, I will explicate a research agenda comprising five tasks for investigating the links between L2 classroom interaction and L2 learners’ developing linguistic repertoires.
2. Overview of usage-based research on language
As previously noted, a usage-based understanding of language is based on converging evidence from research undertaken in several fields. Deriving from this body of research are five fundamentals about the foundations of language (Hall, Reference Hall2019b). First, the composition of language is tied to language use, taking its shape from the ways in which it is used in activity with others within the social contexts of daily life. Second, as individuals participate in their social activities, they draw on a range of emotional and motivational dispositions and cognitive capacities such as perception and categorization that guide individuals to perceive patterns in the use of the resources. Third, key aspects of social experiences that contribute to the development of language knowledge are their recurring nature and ‘the distribution, frequency, and salience of specific linguistic constructions comprising the experiences’ (Hall, Reference Hall2019b, p. 81; see also Bybee, Reference Bybee2006; Goldberg, Reference Goldberg2006). It is not raw frequency in the input that matters but rather the frequency with which linguistic constructions are meaningfully experienced (MacWhinney, Reference MacWhinney2006; Ellis, Reference Ellis and Robinson2012; Crossley, Salsbury, Titak, & McNamara, Reference Crossley, Salsbury, Titak and McNamara2014). All else being equal, the more routine and meaningful learners’ social experiences are and the more frequent, predictable, and salient the linguistic constructions comprising their experiences are, the more likely the constructions will be noticed, conserved, and retrieved as cognitive representations of their experiences.
Fourth, what individuals learn from their extended participation in their social experiences is not a finite, acontexual, abstract system of grammar. Rather, they learn form-function pairings, that is, constructions (Goldberg, Reference Goldberg2006), for making meaning. Constructions include all linguistic phenomena, from single words, to fixed and semi-fixed expressions and collocations or groupings of words. The constructions become stored in learners’ minds as ‘pragmatically driven means for organizing, construing, and experiencing their social worlds’ (Hall, Reference Hall2019b, p. 81). Finally, constructions, once learned, do not remain static in learners’ minds but are continually changing as a consequence of factors ranging from individual attention, motivation, and other individual factors, to ‘competing pragmatic intentions, changing group affiliations, and society-wide forces’ (Hall, Reference Hall2019b, p. 81; see also Lee et al., Reference Lee, Mikesell, Joaquin, Mates and Schumann2009; Douglas Fir Group, 2016). Thus, there is no end state to what is learned. Any appearance of stability in learners’ language knowledge is ‘a matter of stability in their pragmatic pursuits and not an inherent aspect of the resources themselves’ (Hall, Reference Hall2018a, p. 32).
The evidence thus far on the interrelationship between language use and language knowledge is surely compelling. However, it is incomplete in that left unspecified is the consequential role of interactional contexts and their linguistic designs in giving shape to specific linguistic constructions. Growing bodies of research from two interrelated approaches to the study of social life, CA and IL, have been devoted to the study of interaction. Their shared theoretical framework is grounded in the following two premises. First, social order is fundamentally empirical, arising from locally accomplished, recurring sequences of stable, interdependent actions to which participants normatively orient (Sacks, Reference Sacks1992, Reference Sacks1995; Schegloff, Reference Schegloff2007). Second, rather than contained in a single self-organizing system, ‘grammar's integrity and efficacy’ (Schegloff, Ochs, & Thompson, Reference Schegloff, Ochs, Thompson, Ochs, Schegloff and Thompson1996, p. 3) is positionally sensitive, i.e., ‘bound up with its place in larger schemes of human organization and with social interaction in particular’ (p. 3). In their interactions, members of a society respond to each other, negotiate their participation, and, more generally, adapt their positionally sensitive grammars to the contingencies of their interactions as they unfold.
Making this possible is mutually recognizable, publicly observable, interactional infrastructure that members of society use to achieve social order (Heritage, Reference Heritage1984; Sacks, Reference Sacks1992, Reference Sacks1995; Schegloff, Reference Schegloff2007). Rooted in sociology, CA's concern is with describing the components of interactional infrastructure; based in linguistics, IL's concern is with uncovering the recurring linguistic practices of sequentially specific social actions. Together, they add needed empirical weight to usage-based understandings of language by specifying the interactional bases of usage. In Section 3, I briefly overview the contributions of CA and IL to a usage-based understanding of language knowledge and, more specifically, to the proposed research agenda that follows.
3. Contributions of CA and IL to usage-based studies on L2 classroom interaction
CA's research program is concerned with the specification of interactional infrastructure, two structures of which found to be essential to interaction are the systems of turn taking and sequence organization (Schegloff, Reference Schegloff2007; Stivers et al., Reference Stivers, Enfield, Brown, Englert, Hayashi, Heinemann and Levinson2009). The turn taking system involves the construction and distribution of turns in interaction. It comprises two components: the turn-constructional component, which entails how turns are designed, and the turn allocation component, which involves the methods for allocating turns (Sacks, Schegloff, & Jefferson, Reference Sacks, Schegloff and Jefferson1974). Sequence organization has to do with the relative positioning of turns in interaction. Turns are vehicles for accomplishing actions; their positioning in a sequence is fundamental to the analysis of their meanings as actions (Schegloff, Reference Schegloff2007). Sequences or courses of actions are organized not by topical consistency but by actional coherence; actions create meaningful sequences of actions and meaningful sequences of action are responsible for coherence.
The basic unit of sequence organization is the adjacency pair, which consists of two turns that are ‘causally-conditionally related’ (Enfield, Reference Enfield, Stivers, Mondada and Steensig2011, p. 287) such that the initial action projects or makes conditionally relevant a next action. What is considered relevant depends on structural preferences in sequence organization. Structural preferences are the normative expectations about the actional relevance of next turns.Footnote 1 Next turns that lead to the accomplishment of the activity are preferred. Directives, for example, prefer or make conditionally relevant compliant actions (Couper-Kuhlen & Selting, Reference Couper-Kuhlen and Selting2018).
Preference is consequential not only to the types of actions that follow first actions but also to the linguistic designs of the actions. Preferred actions are typically performed more simply or directly than dispreferred actions are, and these differences are displayed in the linguistic practices used to form the actions. Dispreferred actions, for example, are typically longer than preferred actions and formulated with hesitation markers such as ‘uh’ and ‘umm’ and accounts for not performing the preferred actions (Pomerantz & Heritage, Reference Pomerantz, Heritage, Sidnell and Stivers2012). Interlocutors orient to the preference structures of next turns to manage and display to each other their continuing understanding of the work they are doing together.
Sequentially specific actions are the natural environment for the emergence of linguistic constructions (Schegloff, Reference Schegloff, Ochs, Schegloff and Thompson1996; Fox, Reference Fox2007). As noted earlier, constructions are practice-meaning pairings, where practice refers to grammatical forms that are used to carry out actions and meaning is tied to the actions the practices accomplish. An example is the particle ‘oh’ when it is used as a response to some kind of information to register a change of state (Heritage, Reference Heritage and Atkinson1985), as shown in Excerpt 1, line 2. The linguistic item ‘oh’ is the practice and its meaning as a change of state token is its action.
Excerpt 1 (Heritage, Reference Heritage and Atkinson1985, p. 301)Footnote 2
The sequential positioning of a turn and the action it performs shape its grammatical practices, for example, requests in first position are differently formatted than requests occurring later in the sequence (Fox, Reference Fox2007). That is to say, for any particular sequential position in an interaction, there is a restricted set of linguistic practices for carrying out the relevant action in that slot. The term ‘positionally sensitive’ refers to the fact that ‘for any one specifiable sequential position, there is a certain set of forms that can be used to perform the action relevant for that slot’ (Ford, Fox, & Thompson, Reference Ford, Fox, Thompson, Szczepek, Reed and Raymond2013, p. 739). More generally, a positionally sensitive grammar takes grammatical practices to be dynamic, emergent entities, born in interaction and whose meanings are ‘positionally sensitive to organizational features and contingencies of the sequential and interactional moment in which the conduct is situated’ (Schegloff, Reference Schegloff, Ochs, Schegloff and Thompson1996, p. 110).
Practices do not have the same meaning potential across sequential positions. The same practices used in different positions can accomplish different actions, and actions can be accomplished with different practices (Schegloff, Reference Schegloff2007). Take again the particle ‘oh’, for example. As shown above, in Excerpt 1, when used to respond to information giving, it serves as a change of state token. However, when produced in turn-initial positions, it commonly introduces telling sequences and suggests that the speaker has just remembered something he or she wants to tell (Bolden, Reference Bolden2006). This is illustrated in Excerpt 2, line 1.
Excerpt 2 (Bolden, Reference Bolden2006, p. 675)
As another example, the practice of repetitions can be used to display an emotional stance such as surprise or interest or to register receipt of what was just said (Schegloff, Reference Schegloff1997). In classrooms, repetitions can do the work of instructional corrections, but other practices such as explicit negative assessment can also do corrections (Hall, Reference Hall2007). What individuals develop from extended engagement in recurring sequences of action are thoroughly social, positionally sensitive grammars, comprising practice-action constructions linked to the schematic organization of speech situations, and stored as cognitive routines for the accomplishment of social life (Ochs, Schegloff, & Thompson, Reference Ochs, Schegloff, Thompson, Ochs, Schegloff and Thompson1996; Couper-Kuhlen & Selting, Reference Couper-Kuhlen and Selting2018).
Drawing on CA methods and the analytic framework of modern descriptive linguistics (Couper-Kuhlen & Selting, Reference Couper-Kuhlen and Selting2018), IL research investigates the recurrent linguistic and nonverbal practices for accomplishing particular social actions. Social action refers to the main job of a turn, i.e., ‘what the response must deal with in order to count as an adequate next turn’ (Levinson, Reference Levinson, Sidnell and Stivers2012, p. 107). An example of IL research is a set of studies on responsive actions to two types of wh-information-seeking questions: specifying questions, which prefer responses that give particular pieces of information, and telling questions, which prefer extended responses such as reports, tellings, and stories (Fox & Thompson, Reference Fox and Thompson2010; Thompson, Fox, & Couper-Kuhlen, Reference Thompson, Fox and Couper-Kuhlen2015). The research shows systematic differences in the linguistic practices used to perform the two types of information-seeking wh-question-response sequences depending on the work the questions are doing. For example, the expected or preferred response to a specifying question is a phrase. This is shown in Excerpt 3, where Molly asks Felicia a question seeking specific information (line 1) to which Felicia provides the preferred response in the form of a phrase (line 2).
Excerpt 3 (Fox & Thompson, Reference Fox and Thompson2010, p. 134)
Telling questions, on the other hand, prefer extended responses, which typically come in the form of clauses. In Excerpt 4, we see that in response to Vivian's wh-telling question (line 1), Nancy provides an extended report that takes the form of two clauses (line 2).
Excerpt 4 (Fox & Thompson, Reference Fox and Thompson2010, p. 136).
In sum, by evidencing the systematic nature of interaction and the interactional specifications of recurring linguistic constructions in actions and courses of action, together, CA and IL provide explanatory insight into the usage-based sources of linguistic constructions. While CA focuses on what and how actions are accomplished in unfolding sequences of interactional turns, IL's concern is with uncovering the empirical links between sequentially specific social actions and their recurring linguistic designs. Together, their contributions make clear that ‘language knowledge is not only something that originally arose out of usage patterns by speakers, but it is those usage patterns themselves, situated in the interaction itself, as temporal objects representing aggregations of patterns heard and used by the users of the language’ (Laury, Etelämäki, & Couper–Kuhlen, Reference Laury, Etelämäki and Couper–Kuhlen2014, p. 441, emphasis in the original).
A major implication of the research programs of CA and IL for the study of classroom interaction and its links to learners’ L2 development is this: Instructional sequences are the vehicle for accomplishing teaching and learning. To understand more fully L2 teaching actions and their consequential role in shaping learners’ linguistic repertoires requires at the minimum detailed examinations of the actions and action projects along with their linguistic designs comprising naturally-occurring L2 classroom interactions. This is the goal of the research tasks proposed in the next section.
4. The study of L2 classroom interaction and its links to L2 learners’ developing L2 linguistic repertoires
Given the importance of video recordings of naturally-occurring L2 classroom interactions to the proposed research tasks, a brief discussion of data collection protocol is in order. Kimura, Malabarba, and Hall (Reference Kimura, Malabarba and Hall2018) argue that the specialized nature of classrooms requires a particular set of considerations for the production of video recordings of classroom interaction data. Two characteristics of classrooms that call for special considerations are the variety of spatial arrangements that can occur within and across instructional activities and the interplay between sequences of action and larger pedagogical projects.
Regarding the first consideration, they note that L2 classroom contexts typically comprise multiple pedagogical activities within one time period or lesson, which can call for different physical arrangements. For example, instructional activities can be accomplished through lectures, whole-group and small group discussions, group or team projects, pair tasks, and individual work. Each of these can call for a different configuration of space and thus, different placements of recording devices. Knowing ahead of time the types of lessons that are planned can help researchers devise recording plans to ensure maximal coverage of the different spatial arrangements. However, as anyone who has been in a classroom knows, planned activities and their spatial arrangements can change in the middle of a lesson. Thus, researchers must be able to respond quickly to the unanticipated changes and adjust recording devices accordingly.
In terms of the second consideration, the interplay between sequences of action and larger pedagogical projects, Kimura et al. (Reference Kimura, Malabarba and Hall2018) note that sequences of instructional actions are typically organized by larger pedagogical projects. Thus, the meanings of individual actions or sequences of actions may not always be evident within the immediate sequential environment. This has implications for the type and amount of data that is collected, as the length of interaction that is recorded affects ‘the extent to which the analyst is able to depict the impact of institutional demands and overall structural organization’ (p. 14). Thus, to facilitate analysis of the meanings of any one action or sequence of actions and the larger pedagogical goal orientations, researchers should strive to collect recordings of extended sequences of interaction not only within a lesson but also across multiple lessons. Additional considerations when collecting data include the quality of the video recordings, audio backup needs, and so on.Footnote 3
In addition to collecting one's own data, publicly available corpora can be useful sources of classroom interaction data. One such source is the Corpus of English for Academic and Professional Purposes (CEAPP, 2014; ceapp.la.psu.edu). CEAPP is a growing, publicly available, searchable corpus of videos of naturally-occurring classroom interactions and their transcripts. The database includes ClassBank, a corpus of 32 hours of audio and video recordings from various K-12 L1 settings that was originally part of TalkBank.org. CEAPP also includes recordings of university STEM (science, technology, engineering, and math) classroom interaction data, and, relevant for the research tasks here, video recordings from two university-level intensive English as a second language (ESL) grammar courses. To collect the videos from the L2 classrooms, two cameras were used. One captures the teacher view and the other, the student view. Both views are synchronized on CEAPP, making it possible to see what the teacher and students are doing together at all times. CEAPP also contains ethnographic field notes and copies of materials used and/or produced during many of the classroom interactions.
The video recordings are transcribed using the Codes for the Human Analysis of Transcripts (CHAT) system, a standardized format for producing transcripts of face-to-face interactions and which supports the transcription system of CA. The conventions mark the sequential relationships within and between actions including overlaps, gaps, and pauses; the qualities of speech delivery including tempo, pitch, and intonation; and features accompanying talk such as laugh particles, and hesitation markers (MacWhinney, Reference MacWhinney2000).
CEAPP uses the publicly available Computerized Language Analysis Program (CLAN), which provides tools that speed transcription, link transcripts to their videos, and make possible searches for actions and practices that include conversation analytic markings in addition to traditional linguistic units. The videos also make possible the analysis of nonverbal resources such as gestures, gaze, or body posture, that either alone or together with talk, play a role in the organization and design of actions and action sequences. The database is continually being updated with new data and individual users are able to add their data to the database.
To aid in the coding of data, a collaborative commentary component is being developed for CEAPP. This will include the coding protocols produced for research using CEAPP data, such as the study discussed in Research task 1. The protocols will be linked to specific segments of the videos and their transcripts, along with any related materials, and integrated with a platform for users to comment on and provide feedback on the coding protocols and segments (MacWhinney, Reference MacWhinney, Goldman, Pea, Barron and Derry2007). In addition to the independent checking of the codes, this feature will afford data sharing and replication across data sets, researchers and geographical contexts. In so doing, CEAPP will provide ‘a level and public playing-field for the evaluation of competing explanatory accounts’ (Fletcher, Reference Fletcher2014, p. 21).
Research task 1
What are the recurring actions and action sequences comprising the instructional activities of L2 classroom interactions?
As discussed earlier, studies of classroom interaction identified the IRF as a ubiquitous three-action sequence of whole-group instruction, with some critiquing the sequence as a device for limiting students’ participation and others arguing for its usefulness in guiding student learning. The differing views on the instructional value of the IRF are due in part to the limited empirical understandings of teaching actions afforded by the concept. The terms initiation and response, in fact, are not actions, but, instead, are vague glosses for teacher turns that make student turns relevant. The findings to this question will help to identify the specific kinds of actions that turns such as initiations and responses do, which is a first step in understanding the empirical links between teaching and learning.
CA data analysis begins in the qualitative process of transcribing, which requires a set of robust transcription conventions to capture the finely grained details of the simultaneous and sequentially deployed verbal and nonverbal resources used by the teachers and students to achieve mutual understanding in their interactions. In addition to the CLAN conventions, recommended transcription conventions are Jefferson's (Reference Jefferson and Lerner2004) conventions for linguistic and paralinguistic conduct and Mondada's (Reference Mondada2018) conventions for nonverbal conduct.
The next steps entail identifying and coding the actions and action sequences and gathering them into collections. These steps call for the use of analytic induction and constant comparison methods (Heritage, Reference Heritage and Silverman2010; Sidnell, Reference Sidnell, Sidnell and Stivers2012; Couper-Kuhlen & Selting, Reference Couper-Kuhlen and Selting2018). Determining the actions of turns involves consideration of their linguistic designs, i.e., their prosodic and grammatical practices, and sequential practices such as overlaps of turns and gaps of silence within and between turns. In addition, it entails consideration of the sequential positionings of the turns in relation to prior turns, to how the turn is responded to in the next turn and their positions in sequences of actions in relation to the larger ongoing activity (Sacks et al., Reference Sacks, Schegloff and Jefferson1974; Enfield & Sidnell, Reference Enfield and Sidnell2017).
Two key aspects of CA make it the most quantitative of the qualitative social science methods (Stivers, Reference Stivers2015). First, classifying actions and their practices requires ‘a clear characterization of what “counts” as an instance of the target phenomenon’ (p. 3). Thus, the focus is not on the distinctiveness of each case but on their shared properties. Second, it seeks distributional evidence by studying the patterning of actional properties within and across data sets. A procedure used in CA to check the validity and generality of analyses is deviant case analysis (Edwards, Reference Edwards and Jupp2006). Deviant cases are those that do not appear to follow the proposed pattern. They can help in revisions and confirmations of patterns emerging from data analysis.
Two methodological challenges of coding are worth noting. First, analysts need to be mindful when assigning actions to practices as there is no ‘one-to-one practice/action pairing’ (Schegloff, Reference Schegloff1997, p. 505). As noted earlier, linguistic practices such as repetitions, for example, can be used to take different actions. A teacher's repetition of the student utterance ‘He didn't come home’ with rising intonation can serve to seek a correction of the utterance, to express surprise at what was stated, or to initiate a repair due to problems in hearing or understanding the utterance (Hall, Reference Hall2007). Similarly, teacher questions can serve different actions. They can seek information, make a request, assess a prior turn, and so on. The actions of questions seeking information can also vary depending on whether they are seeking specific or telling information, as discussed earlier (Fox & Thompson, Reference Fox and Thompson2010; Thompson et al., Reference Thompson, Fox and Couper-Kuhlen2015). One must strive, then, ‘to remain alert to an action-formation resource pool, in which practices, deployed always in some position, can accomplish different actions; and actions can be accomplished through a variety of situated practices’ (Schegloff, Reference Schegloff1997, p. 505, emphasis in the original). Second, while the actional meanings of many utterances may be clear, analysts must be alert to the possibility of ambiguity when coding. The uncertainty can be due to the analyst's lack of ethnographic knowledge to which the individuals in the interaction are orienting or because the ambiguity is internal to the data, for as Schegloff (Reference Schegloff1997, p. 520, emphasis in the original) noted, individuals ‘do speak in ambiguous ways.’ Thus, it is advised to add ‘ambiguous’ as a coding category to ensure nuanced analyses.
An example of a study seeking to uncover the recurring actions and action sequences comprising the instructional activities of L2 classroom interactions is the ongoing study by my colleagues and me of teacher information-seeking questions (Hall, Khor, & Wang, Reference Hall, Khor and Wang2019; Khor, Hall, & Wang, Reference Khor, Hall and Wang2019). Based on our analyses of four hours of data from an ESL classroom included in CEAPP, we found that over 90% of teacher questions that occurred over the four hours were information-seeking questions, and of these, almost 65% were specifying question actions that called for specific information responsive actions.
Excerpt 5, taken from CEAPP, is an example of such a sequence of action. The sequence takes place during an activity in which the class is discussing a short clip taken from the TV series The Big Bang Theory. Here we see JUL, the teacher, asking a wh-fill-in-the-blank information specifying question (line 1) to which the student provides a specifying response (line 3).
Excerpt 5 (CEAPP, 2014; GR4J012213T6Footnote 4)
Telling question actions that called for telling responsive actions comprised 35% of the total number of information-seeking questions. Excerpt 6 is an example of this sequence of action. It also takes place during the discussion of the clip from the TV series. Here, JUL asks a so-prefaced wh-question (line 1), which elicits a telling responsive action from the student (line 3).
Excerpt 6 (CEAPP, 2014; GR4J012213T6)
In our ongoing analysis, we anticipate finding that teacher questions perform a range of actions including requesting, directing, assessing, seeking agreement, and disagreeing, in addition to seeking information, and that types of question actions are consequential to the types of actions of student responses. We also anticipate finding that larger sequences of teacher-student action sequences vary according to the instructional activity they accomplish. The study of the recurring actions and action sequences comprising the instructional activities of L2 classroom interactions can easily be extended to other L2 classroom contexts.
Research task 2
What are the linguistic patterns of the recurring sequentially fitted actions?
Once the actions and action sequences comprising instructional activities have been identified, the next step is to identify their linguistic designs and the patterns that occur between the designs of teacher actions and those of student responsive actions. The study by Thompson et al. (Reference Thompson, Fox and Couper-Kuhlen2015) of the grammar of responsive actions provides a useful example of how to go about this. They examined four types of sequences: responses to two types of wh-questions, responses to informings, responses to assessments, and responses to requests; and six linguistic designs of responses: particle, lexical/ phrasal, minimal clausal, expanded clausal, gradable clausal, which are responses that are upgraded or downgraded with respect to the initiating action, and unrelated clausal, which do not use any linguistic practices from the initial action. Thompson et al.'s findings show how responses to the different initial actions are designed with different grammatical practices depending on speakers’ motivations.
Examples of responses to the two different types of information-seeking questions were presented in Section 3. As another example, the researchers found that particle responses such as okay and sure to requests indicate voluntary compliance on the part of the requestee to completing the requests whereas the minimal clausal response I will, while accepting of the terms of a request, is a more assertive response that indicates the requestee's agency and will to perform.
Drawing on the work of Fox and Thompson (Reference Fox and Thompson2010) and Thompson et al. (Reference Thompson, Fox and Couper-Kuhlen2015), my colleagues and I (Hall et al., Reference Hall, Khor and Wang2019; Khor et al., Reference Khor, Hall and Wang2019) examined the linguistic designs of both specifying and telling information-seeking questions in the four hours of L2 classroom interaction data detailed for Research task 1, above. and those of the student responsive actions to the questions. We found that specifying questions were designed primarily as yes/no interrogatives (YNI) and wh-questions, at 38% and 36% respectively. Excerpt 5 illustrates a wh-question designed as a fill-in-the-blank question. Excerpt 7 is an example of a YNI specifying question action (lines 1 and 2).
Excerpt 7 (CEAPP, 2014; GR4J011713T4)
Additionally, we found that 90% of student responses to specifying YNIs are designed as specifying one-word or multi-unit phrasal responses. Examples of one-word responsive actions can be seen in Excerpts 5 and 7, lines 3 and 4 respectively. Additional designs of specifying questions include alternative either/or questions and decidedly incomplete utterances (Koshik, Reference Koshik2002), in which the teacher begins reciting the answer she is seeking, stops before the end, and leaves the remainder to be completed by the responder.
In terms of telling questions, we found that 74% are designed as wh-questions and that these types of questions engender telling responses in the form of simple clauses. The student response in Excerpt 6 is an example of such a response. Moreover, while we have found that, generally, specifying questions engender specifying responses and telling questions engender telling responses, there are occasions when specifying questions yield telling responses and telling questions elicit specifying responses. We suspect that the larger instructional activity plays a significant role in eliciting the different responses. Ongoing analyses will provide greater insight into the relationship between the instructional activity and information question sequences. For now, we can state with confidence that using the methods of IL gives us greater purchase on the links between teaching and learners’ developing L2 linguistic repertoires in ways that the concept of the IRF does not. The study of the linguistic designs of the recurring actions and action sequences comprising L2 classroom instructional activities can easily be extended to other L2 classroom contexts.
Research task 3
What are the longitudinal effects of the linguistic patterns of the recurring action sequences on L2 learners’ language use and their developing L2 linguistic repertoires?
Once the linguistic designs of the actions and action sequences comprising instructional activities have been identified, the next step is to investigate the effect of the actions and their linguistic designs on learners’ developing linguistic repertoires. Microgenetic studies are ideally suited for this (Siegler, Reference Siegler, Kuhn, Siegler, Damon and Lerner2006). They are small-scale studies in which the linguistic behaviors of an individual or small group are observed and recorded over a period of time to identify changes, and the specific contextual conditions of these changes. The value of microgenetic studies is that they allow researchers to see not only whether change in learners’ language use occurs but also to link specific changes to specific contextual conditions. Moreover, they allow researchers to identify whether change ‘may be sudden and discrete, or smooth and gradual and, importantly, whether different individuals experience the same transitions’ (Flynn, Pine, & Lewis, Reference Flynn, Pine and Lewis2007, p. 4). Thus, they allow researchers to understand changes in behavior in a way that more traditional methods cannot.
The methods entail intensive video-recorded collections of individuals’ actions over a specified period of time, along with collections of other visual, written, or electronic records, and additional ethnographic data as relevant. The density of recordings should be high enough to capture even the smallest changes in learners’ language use. Qualitative and quantitative measures are then used to analyze the data. Qualitative analyses can uncover the changing social actions and their linguistic designs as they happen in response to teacher actions. Descriptive and inferential quantitative measures, such as frequency counts, sign tests, and other nonparametric measures, can also be used to detect whether any changes in the learner actions and their linguistic designs are significant from one point to another, and how the changes relate to specific teacher actions and larger action sequences.
According to Siegler (Reference Siegler, Kuhn, Siegler, Damon and Lerner2006), microgenetic studies address five dimensions of change. The source of change has to do with the events that facilitate the change. The path of change is the particular type of change, whether it is qualitative, resulting in different types of knowledge or ability, or quantitative, resulting in changes in speed or accuracy, or whether both types of change occur. The rate of change refers to how quickly the change occurs. The breadth of change has to do with whether the change is domain-specific or applicable to several domains and the variability of change has to do with how consistent the patterns of changes are across time, across contexts, and across individuals.
Kim and Hall (Reference Kim and Hall2002) and van Compernelle (Reference van Compernelle2010) are two studies that have used microgenetic methods to study L2 learning. Kim and Hall examined connections between the participation of a small group of native-speaking Korean children in an interactive book-reading program and their development of particular linguistic and interactional practices in English. Van Compernelle used microgenetic methods to examine the collaborative construction of an object of learning between a learner of French and a teacher during an oral proficiency interview.
Drawing on the methods used in these two studies and using the video-recorded data collected for the first two research tasks, one can track the actions and their linguistic designs of a student or group of students within and across instructional activities to determine the five dimensions of change. Examining the sources of changes to learner language can help determine whether the change is tied to particular types of teacher actions and whether these actions are tied to particular types of instructional activities. The paths and rates of change can be determined by examining in close detail the linguistic and interactional designs of learner actions, looking for changes in, for example, the timing of student actions, the use of prosodic and other paralinguistic and nonverbal cues in the designs of their actions in response to teacher actions. The breadth and variability of change can be determined by examining whether the changes are specific to particular actions and action sequences, and more generally to types of instructional activities.
Research task 4
What impact does the redesign of teaching actions and their linguistic practices have on L2 learners’ language use and their developing L2 linguistic repertoires?
There is a history of using CA studies revealing the linguistic and interactional intricacies of interactions to inform and assess the success of communication interventions in various institutional settings, including medicine, business, and counseling clinics (Heritage, Robinson, Elliot, Beckett, & Wilkes, Reference Heritage, Robinson, Elliot, Beckett and Wilkes2007; Kitzinger, Reference Kitzinger and Antaki2011; Stokoe, Reference Stokoe and Antaki2011, Reference Stokoe2014; Robinson & Heritage, Reference Robinson and Heritage2015). One example of an interventionist CA study is that by Heritage and his colleagues (Heritage et al., Reference Heritage, Robinson, Elliot, Beckett and Wilkes2007; Robinson & Heritage, Reference Robinson and Heritage2015) of doctor-patient acute-care visits. They noted that while doctors’ full knowledge of their patients’ concerns in these visits facilitates diagnosis and treatment, they often face difficulties in effectively eliciting patients’ concerns. Based on their observations that doctors’ questions affect the substance of patients’ responses about their concerns, Heritage and his colleagues designed a study using two different question designs as in ‘Is there anything else you want to address in the visit today?’ and ‘Is there something else you want to address in the visit today?’ They speculated that because the negative polarity of ‘any’ prefers a ‘no’ response, the question using ‘some’ would elicit more concerns from patients than the question using ‘any’ would. This was borne out in their findings in that the use of ‘some’ elicited more concerns from the patient than the ‘any’ question did. They concluded that the ‘some’ intervention significantly reduced the probability that patients left their visits with the doctors with unmet concerns.
Like other institutional contexts, such as doctor-patient consultations, classrooms are natural laboratories in that their interactions can be engineered to produce different kinds of instructional actions that engender different kinds of learner actions and more generally to increase the range of interactional opportunities for L2 learners and the linguistic quality of their actions. The study on the use of ‘some’ versus ‘any’ offers an interesting model for the design of an intervention study for the classroom. In our preliminary study of information-seeking questions (Hall et al., Reference Hall, Khor and Wang2019; Khor et al., Reference Khor, Hall and Wang2019), we found that the question ‘anything else?’ elicited no responses from students twice as often as the question ‘what else?’ did. We speculate that this is due to the negative polarity of the use of ‘any’, which, as noted above, prefers a ‘no’ or no response.
We plan to further investigate whether there are, in fact, any significant differences between the two questions by searching the CEAPP data and conducting quantitative analyses using nonparametric and multivariate measures. One can also design a small-scale quasi-experimental intervention study with L2 teachers to test whether the question ‘what else?’ produces more student responses during whole-group instruction than the question ‘anything ‘else?’ does. Knowing this would help teachers and teacher educators understand how their teaching actions are linked to opportunities for learners to engage in interaction. Once interested teachers are identified, they can be placed into two groups. To one group, the goal of the study and the intervention of ‘what else?’ versus ‘anything else?’ should be described, explained and modeled. The other group should receive no special instructions. Since the focus is on the number of questions and not the number of teachers, the total number of participants can be fairly small but the hours of video recordings should be substantial enough to capture enough information-seeking questions to distinguish whether there are differences between the two focal questions in terms of student responses and whether the differences are significant.
Another designed-based intervention study can draw on the findings from the study of the linguistic designs of student responsive actions to specifying and telling information-seeking questions, summarized in Research tasks 1 and 2. As noted, we found a preponderance of specifying question actions versus telling question actions and student responses that were mainly phrasal responses. To increase the possibility of longer student responses that are more linguistically complex, L2 teachers can be recruited to participate in the redesign of instructional activities to include more telling question actions, and then to use the redesigned activities as part of an intervention study to determine the impact of the redesigned activities on L2 learner language use.
Research task 5
How does L2 teacher use of CEAPP, an online video corpus of L2 classrooms, enhance their understandings of their role as key designers of L2 classroom learning environments and L2 learners’ linguistic repertoires?
In addition to their use in intervention studies, CA methods and findings on institutional interactions have been used to train professionals in workplace settings by enhancing their understandings of the specialized communication skills needed in their settings (Antaki, Reference Antaki and Antaki2011; Finlay, Walton, & Antaki, Reference Finlay, Walton, Antaki and Antaki2011; Stokoe, Reference Stokoe and Antaki2011, Reference Stokoe2014; Hepburn, Wilkinson, & Butler, Reference Hepburn, Wilkinson and Butler2014). The general focus of the approach is on working collaboratively with members of the focal institutions to consider the findings on interactions associated with particular institutional activities and roles, and, based on the considerations, develop possible solutions to any identified problems.
One approach that has demonstrated success in training and assessing the specialized communication skills of professionals is CARM, the conversation analytic role-play method (Stokoe, Reference Stokoe and Antaki2011, Reference Stokoe2014). CARM provides participants with ‘a unique opportunity to examine communicative practices in forensic detail, and to understand what works from a rigorous empirical basis’ (Stokoe, Reference Stokoe2014, pp. 1–2) by using video extracts of real-life interactions from the workplace settings of participants as the basis for role-play and discussion. The videos include both moments of interactional troubles and successful outcomes. As participants watch the videos, the recordings are stopped at certain points so that the participants can discuss the ways in which the interactions appear to be helping or hindering anticipated outcomes and role play possible solutions. They then watch and evaluate how the issues are actually dealt with.
The use of CA methods and adaptations of the CARM approach to improve L2 teacher training is on the rise (see Glaser, Kupetz, & You, Reference Glaser, Kupetz and You2019, for a discussion on the use of CA in L2 teacher training). An early application is by Sert (Reference Sert2013) who used methods similar to CARM in an L2 language teacher education program to improve classroom interaction practices. In Sert's study, L2 teachers were asked to identify and reflect on student claims of insufficient knowledge such as ‘I don't know’ as they viewed excerpts of video recordings from an L2 classroom. They found that, in many cases after students claimed a lack of knowledge, teachers often moved on to other students to provide the sought-after answers. The teachers then considered alternative actions that would engage the students who claimed a lack of knowledge and lead to demonstrations of understanding.
I have adapted methods of the CARM approach to explore L2 teaching and learning as displayed in the videos on CEAPP in workshops given in methods courses for pre-service L2 teachers. Turn allocation and student displays of engagement are two of the many topics we have explored. Before examining the videos, students are assigned readings on the topics. In the case of turn allocation and student engagement, for example, they are asked to read Sahlström (Reference Sahlström2002) on the morphology of student hand raises and Kääntä (Reference Kääntä2012) on teachers’ use of gaze, head nods, and pointing gestures to allocate turns to students. After discussing the readings in class, the students are then directed to video clips on CEAPP and, in pairs, are asked to watch them and take notes on how turns are allocated and on how students display engagement and disengagement. In a large group, we then discuss their observations and consider their experiences as students in terms of turn allocations and engagement and discuss their reflections on why students might disengage and steps that teachers can take to increase their engagement. The students in the methods course have reported on their end-of-semester evaluations that the workshops on teaching using CEAPP have a major positive impact on their understandings of the specialized nature of teaching.
For this research task, the methods used in the CARM approach and the findings on the two types of teacher questions and the types of student responses they engender can be drawn on to create a workshop for pre-service and in-service L2 teachers that uses the data on CEAPP similar to the workshop described above. The aim is to raise their awareness of their role as key designers of classroom interactions. The first step entails identifying a set of excerpts that feature the two types of information question action sequences. Participants are then led through the teacher questions of each excerpt. Then the video is stopped and the participants are asked to discuss possible student responses to the question in terms of what they would consider to be effective responses. Once the participants have discussed possible responses, the video is continued so that they can see and compare what they think should happen and how the students actually respond. This can then lead to a discussion of how the teacher questions could be redesigned to elicit longer and more complex student responses. To determine the teachers’ perceptions of the value of using CEAPP data to enhance their understandings of teaching and of perceived changes to their instructional skill sets, they can be asked to complete a questionnaire or evaluation.
5. Final thoughts
It is a fact that interactions between L2 teachers and students are consequential to L2 learners’ developing linguistic repertoires. The intentional choices that L2 teachers make in designing the actions and action projects of their instructional activities give shape not only to linguistic resources that are made available to learners in their activities but as importantly ‘to the ways in which L2 learners orient to the resources and use them’ (Hall, Reference Hall2019b, p. 90). The aim of the research agenda presented here is to increase L2 teachers’ and other stakeholders’ understandings of the sophisticated, complex work of teaching so that they can redesign their classrooms in ways that make possible new ways of teaching and learning. For as Heap (Reference Heap1990, p. 43) notes, ‘if some activity is important in our lives, then knowing how it is organized may make a difference to how we act’.
Traditionally, teachers have worked independently. However, developing sustainable solutions for the real-world challenges of L2 teaching require collaborative teams of teachers to engage in these tasks, working interdependently to reflect on their teaching practices, derive new insights and understandings about teaching, and make positive sustained changes to their teaching practices (Hall, Reference Hall, Haneda and Nassiji2018). Such work will make significant contributions to the advancement of understandings of second language acquisition and, specifically, the transformative role of L2 teaching to learners’ developing linguistic repertoires.
Joan Kelly Hall is Professor of Applied Linguistics and Director of the Center for Research on English Language Learning and Teaching (CRELLT) at the Pennsylvania State University. Her research centers on documenting the specialized interactional practices and actions of teaching-and-learning found in instructional settings. Her work appears in journals such as Applied Linguistics, Journal of Pragmatics, and Modern Language Journal. Her most recent books are Essentials of SLA for L2 teachers: A transdisciplinary framework (2018, Routledge) and The embodied work of teaching. Multilingual (with S. D. Looney, 2019, Multilingual Matters).
