2.1 Research projects focused on fluency and pronunciation

While ITA educators remain interested in all aspects of pronunciation (Meyers & Holt Reference Meyers and Holt2001; Gorsuch Reference Gorsuch and Gorsuch2012a), suprasegmental aspects of pronunciation have gotten some focused attention, which has been expressed in research and materials (see also Isaacs (Reference Isaacs2008) for pronunciation research conducted with an ITA population). There have been descriptive studies based on ITA learner populations on sentence-level stress (Hahn Reference Hahn2004; Levis, Muller Levis & Slater Reference Levis, Muller Levis, Slater and Gorsuch2012), and tone choices and intonation (Wennerstrom Reference Wennerstrom1998; Pickering Reference Pickering2001) with an aim to linking ITAs’ spoken intelligibility to these aspects of pronunciation. There has been pedagogical intervention research on ITAs’ pausing patterns (pauses violating clause and sentence boundaries) (Gorsuch Reference Gorsuch2011a; Gorsuch Reference Gorsuch2013) motivated by descriptive studies linking pausing patterns to listener perceptions of spoken fluency (e.g., Ejzenberg Reference Ejzenberg and Riggenbach2000; Olynack, Anglejan & Sankoff Reference Olynak, Anglejan, Sankoff, Scarcella, Andersen and Krashen1990; Pawley & Syder Reference Pawley, Syder and Riggenbach2000).

Pickering (Reference Pickering1999, see also Reference Pickering and Malmkær2010) applied a then-untested descriptive framework to ITA learner populations (Brazil Reference Brazil1997) with explicit linkages between language, teaching, and culture. For instance, in extended speech, a speaker of English uses prominence (sentence-level stress) to mark new or important information (which teachers need to do). English speakers also use tone choices (rising, level, or falling intonation contours) to express what the speaker believes is new or known information (which teachers need to do), and to express engagement and authority (which teachers also need to do). A textbook for ITAs based on this descriptive framework was published (Gorsuch et al. Reference Gorsuch, Meyers, Pickering and Griffee2013), and other books on pronunciation and academic discourse for advanced-level learners, have appeared (e.g., Hahn & Dickerson Reference Hahn and Dickerson1999). Interestingly, even the availability of principled, pedagogical presentations of suprasegmental aspects of pronunciation has failed to result in suprasegmentals becoming anything like an organizing force in ITA program curricula (Hahn Reference Hahn2004; Gorsuch Reference Gorsuch and Gorsuch2012a). This review sets the stage for two projects.

The two studies mentioned here firmly linked ITAs’ sentence-level stress and holistic fluency measures to undergraduates’ comprehension and recall of extended academic talk (cf., Field (Reference Field2005) and Kang (Reference Kang2010)). Using large groups of undergraduates (N = 224 and N = 90, respectively), Anderson-Hsieh & Koehler and Hahn built up a consensus base of listeners’ perceptions connected directly to content learning. Yet this has not really registered with ITA educators. Carrying out additional studies will do more to convince current ITA educators of how salient ITAs’ needs are for appropriately using just a few well-defined suprasegmental aspects of pronunciation in academic talk. Thus far, only male speakers with first language (L1) Mandarin Chinese or Korean have comprised the ITA speakers. To be fair, the two studies explored variables such as speech rate and sentence stress patterns on undergraduate comprehension, and thus the researchers controlled for speakers’ L1s in favor of variables of greater interest to them. Nonetheless, ITAs are a diverse group, being equally male and female, and coming from dozens of countries and regions including Turkey, Jordan, the African continent, Brazil, and the Indian sub-continent. An ITA speaker pool for research linking ITAs’ use of suprasegmentals to undergraduate comprehension and learning should take into account this diversity, allowing variables involving ITAs consciously speaking faster or slower to the undergraduate group to take a back seat. Using a more diverse speaker pool may convince ITA educators of the contribution of well-defined features of continuous speech to ITA intelligibility across L1s.

Both Anderson-Hsieh & Koehler and Hahn employed large groups of native English speaker listeners, which I think was possible because the number of ITA speech samples they had to listen to was limited. For example, Hahn (Reference Hahn2004) used one male Korean speaker reading the same script but with different sentence stress patterns. Anderson-Hsieh & Koehler used three speakers with three different English proficiency levels. If the number of speech samples a group of learners has to listen to is greater in number, it may be more difficult to work with large, intact undergraduate groups. A ‘larger speech sample’ might be defined as three to five L1s, and perhaps non-fluent and fluent speaker groups speaking on the same topic for each L1. Smaller groups of 30 NS listeners might do, where undergraduates could come in independently by appointment to one or two dedicated computers with the multiple speech files, perception questionnaires, and content recall quizzes ready to use.

There has been accumulating evidence of how ITAs use, and do not use, key aspects of suprasegmental pronunciation in comparison with native English speaking TAs. This has significant implications for ITAs’ success in teaching, and their ability to negotiate cultural norms in U.S. classrooms (for a review, see Gorsuch Reference Gorsuch2011b). In other words, we need to move beyond the sentence-level stress feature that Hahn (Reference Hahn2004) focused on, and the holistic fluency measure Anderson-Hsieh & Koehler used. For instance, Pickering (Reference Pickering1999: 51‒52) found that ‘pauses in . . .NNS data were both longer and more erratic. . .and tended to regularly break up conceptual units,’ meaning that their pauses violated clause boundaries. Working from two different descriptive frameworks, both Wennerstrom (Reference Wennerstrom1998) and Pickering (Reference Pickering1999) found that ITAs used a preponderance of level and falling intonation contours, which had the potential of making the ITAs sound disengaged or unfriendly. These same researchers primarily used speech analysis software (e.g., ‘machine measures’) using precisely measured pause times, pitch frequencies, and volume measurements in decibels to reach their conclusions. In order to create measures that were less dependent on advanced knowledge of sophisticated speech software and database construction, Gorsuch (Reference Gorsuch2013) used basic features of speech analysis software to identify pauses, sentence stress, and pause group final pitch as a starting point, but then also adapted and devised per-minute measures for these areas of speech.

The point here is that machine measures, regardless of how they are done, should be shown to have consistent relationships with listener perceptions and more importantly, listener comprehension of ITAs’ extended speech. In other words, listener perceptions should be quantified as specific measures of suprasegmental aspects of pronunciation that are meaningful to a general audience of ITA educators. There would be at least two benefits. First would be to learn whether, and which, machine measures of suprasegmental aspects of pronunciation had the strongest, or most consistent, relationships with listener perception and comprehension and recall of content. This could help to prioritize which aspects of pronunciation ought to be focused on in the classroom. Second, it would also serve to underscore the salient role suprasegmental aspects of pronunciation has for ITA classroom communication success and undergraduate learning, and may result in a more principled treatment of these areas of speech in ITA training.

The developmental path of spoken fluency and prosody has yet to be well described. I use the perhaps more familiar term, ‘prosody’, to mean rhythm, stress, and intonation, also described in the accounts of discourse intonation outlined earlier. There is some conceptual overlap between fluency and prosody, in that listeners’ perceptions of fluency depend in part on whether a speaker's utterances occur as complete idea units with an accompanying intonation contour (see previous discussion). Some broad developmental stages have been suggested where second language (L2) learners either have fluency and prosody or do not have it (e.g., Wennerstrom Reference Wennerstrom1998, Reference Wennerstrom2001; Pickering Reference Pickering1999). The only other real, and depressing, finding is that there is much evidence for idiosyncracy in acquiring fluency and prosody (Freed Reference Freed and Freed1995; Wennerstrom Reference Wennerstrom2001).

Measures of fluency (such as pausing and speech rate) and prosody need to be done longitudinally on a diverse cohort of ITAs (L1s, levels, gender, etc.) spanning at least a two-year period. The measures should be based on listener perception and machine measures, and the ITAs need to be asked to do at least three parallel tasks at each data collection point: a read-aloud task, an extemporaneous one-on-one free-speaking task, and a teaching simulation task with some preparation time allowed. Having the longitudinal measures would establish which aspects of speech change over time, and how. There may be perhaps much variation on some feature at one point in time with less variation later, or more ‘progress’ at the beginning on some feature but not later on, etc. Having three different tasks would give an idea of what type of knowledge ITAs are using to do a given task, which could provide a window into how fluency and prosody are developed. For instance, on a read-aloud task, ITAs with poor initial fluency but with some formal instruction in fluency and prosody may do quite well because they are using explicit knowledge. The language processing demands on a read-aloud would be low compared to a free-speaking task. Such learners would not do as well on free-speaking tasks or teaching simulation tasks because processing demands increase in those conditions and learners could not invoke their explicit knowledge. However, ITAs with more implicit knowledge, or more proceduralized explicit knowledge, might do better later on, across all three tasks.

Such developmental accounts need to include detailed ITAs’ histories: what formal ESL/ITA classes they had and when, what mentoring in teaching they had, what teaching they have been doing and how much, what they believe their level of intercultural competence is at a given point (e.g., LaRocco Reference LaRocco and Gorsuch2012), and whether they participate extensively in L1 social groups outside of immediate academic life. It is true that many other variables than those suggested here may account for ITAs’ fluency and prosody development. But such a study as proposed here may capture patterns that can be used to build and test models, not just for ITAs, but for other advanced-level learners. Results from this project would inform ITA educators about: (1) What kinds of instruction, such as explicit- and implicit-knowledge building, are effective; (2) How ITAs’ personal and professional history of L2 use influences their L2 growth; and (3) How long ITAs’ fluency would take to develop to a point they were considered successful classroom communicators.

2.2 Research projects conjoining ITAs’ L2 development to teaching or culture

I think that curricular responses which consider the components of ‘language, teaching, and culture’ in isolation may have questionable value. If culture is divorced from language, for instance, an ITA program might be reduced to presenting ‘norms’ of undergraduate behavior to ITAs through reading assignments or lectures on ‘what to expect from North American undergraduates.’ This would not support ITAs’ L2 growth (Gorsuch Reference Gorsuch and Gorsuch2012a): First, the texts or lectures are not processed as L2 input and second, such an approach ignores an ITA program-relevant perspective that culture, and teaching is instantiated in verbal, social interactions (Gumperz & Cook-Gumperz Reference Gumperz, Cook-Gumperz, Kotthoff and Spencer-Oatey2007). Various commentators have argued for more emphasis on teaching (e.g., LoCastro & Tapper Reference LoCastro and Tapper2006; Kim & Kubota Reference Kim and Kubota2012) or culture (e.g., Kang & Rubin Reference Kang and Rubin2012) for ITA support, partly out of a resistance to viewing ITAs as linguistically deficient. Yet L2 ability and use is a pervasive subtext that cannot be discounted. For ITAs to learn how to teach, or access teaching experience gained in their home countries, they have to develop procedural knowledge, or ‘how to perform things’ in their L2, which is English (Chiang Reference Chiang2009: 464; Gorsuch Reference Gorsuch2012b). Further, as Kim (Reference Kim2001: 36) argues, one main means of cultural adaptation ‘occurs in and through communication,’ which is, again, in English, the L2.

At the same time, there is some sentiment among ITA educators that ITA programs are not designed to develop ITAs’ L2 vocabulary and grammar (‘language’). Rather, programs should seek to help ITAs learn to better use what L2 they already ‘have’ (Gorsuch Reference Gorsuch and Gorsuch2012a). Disclaiming ‘language’ as a major program goal, and replacing it with ‘good teaching’ are together seen as helping ITA programs avoid being labeled as ‘remedial’ (Gorsuch Reference Gorsuch and Gorsuch2012a), which is seen as politically and institutionally disadvantageous. ITA programs have favored spoken language production along these lines: do spoken task, preferably a teaching simulation ⇒ get feedback ⇒ do spoken task again (e.g., Byrd & Constantinides Reference Byrd, Constantinides and Byrd1995; Alsberg Reference Alsberg, Davis, Smith and Smith2002; Papajohn et al. Reference Papajohn, Alsberg, Bair, Willenbrough and Orr2002; Gorsuch Reference Gorsuch and Gorsuch2012a). At the same time, programs have somewhat avoided systematic attention to L2 grammar and vocabulary growth, either through listening or reading treated as L2 input, or through substantive explicit instruction (Gorsuch Reference Gorsuch2011a).

Levis et al. (Reference Levis, Muller Levis, Slater and Gorsuch2012) demonstrated convincingly that ITAs do require help with their L2 as teachers of freshman-level content (conjoining ‘language’ and ‘teaching’). The researchers had Indian, Chinese, and American engineering TAs with at least two semesters of teaching experience give a presentation based on a silent reading of a physics text on the concept energy, plus preparation time. This is a common teaching simulation preparation technique in ITA education (J. Levis, personal communication, October 1, 2013). Using a Systematic Functional Linguistic descriptive framework, they found that the three groups of TAs structured their oral explanations with varying degrees of ‘success,’ meaning conceptual clarity and internal coherence. Compared to American TAs, the ITAs used lexical and grammatical choices that in the main misrepresented or under-conceptualized the text content. Levis et al. (Reference Levis, Muller Levis, Slater and Gorsuch2012) described specific types of choices that contributed to a lack of cohesion in the information structure, and attributed these lexical and grammatical choices to ITAs’ lack of L2 resources. The fact that the ITAs had already been at the university for two semesters emphasizes that simply being in the U.S.A. does not guarantee critically needed L2 growth (Gorsuch Reference Gorsuch2011a), and that ITA education as a field needs to find testable methods for building ITAs’ repertoire of lexis and grammar specific to their work.

Levis et al.'s (Reference Levis, Muller Levis, Slater and Gorsuch2012) findings lay smack on top of a contradiction that ITA educators get caught up in. We work with advanced L2 learners who work in academic areas we are not completely familiar with, such as chemistry or physics. Yet we must still be experts on how to communicate in English about concepts from these fields. Our practice of having ITAs read a basic text from their field and then base a teaching simulation on it may be a result of this perceived contradiction, hence our concern that ITAs are able to teach concepts North American freshman students are expected to learn. This is also why we use texts on foundational concepts. But we also may take these texts as exemplars of how North American academe believes concepts should be presented, and how undergraduates expect content to be presented. These texts may thus form the basis of a testable method for ITAs to learn L2 lexis and grammar relevant to their professional needs: that is, to give coherent explanations conducive to undergraduate learning. As noted above, the research task suggested here lies within a teaching simulation preparation model already familiar to ITA educators.

Levis et al.'s (Reference Levis, Muller Levis, Slater and Gorsuch2012: 541‒542) teaching simulation preparation task consisted of:

1. 1. Having ITAs silently reading a 130-word text which comprised a basic definition of the concept ‘energy’

2. 2. Having ITAs prepare a teaching presentation for 20 minutes

3. 3. Having ITAs give a 4- to 6-minute talk based on the text

In the study, ITAs’ presentations were video-recorded and transcribed. To develop this basic task into a course of treatments, I suggest finding texts of similar length and audience, such as basic freshman texts, which also exemplify established ways of structuring information. Levis et al. identified one such structure: ‘[a] logical taxonomy of items, and explained and related to each other in some way. . .the speaker can [then] sum up the key relational points s/he wants the audience to remember’ (2012: 550). ITAs’ own academic disciplines are to some extent engaged in identifying these structures to be used in professional talks (see, for example, Leonhard Center for the Enhancement of Engineering Education 2013).

Different researchers/educators could use different instructional approaches to exploit the video or audio recordings of the teaching simulation resulting from the basic task. Here is one: Have ITAs transcribe their own video files and those of their classmates through a repeated listening and dictogloss procedure. This is a procedure where small groups of learners jointly reconstruct the talk (Wajnryb Reference Wajnryb1990; Vasiljevic Reference Vasiljevic2010). This would increase ITAs’ level of attention to the grammatical and lexical features in a given presentation. This hypothesis could be tested through observations or recordings of ITAs’ interactions, or stimulated recall post-reconstruction interviews, using the resulting transcription as the stimulus. Then, ITAs could be lead through a guided discussion of the information structure of the original text passage with explicit comparisons to the information structure of an oral presentation, with the aim of encouraging the class and the ITA educator to jointly generate alternative grammatical and lexical choices that attain adequate conceptual clarity and coherence. If ITAs themselves were given increasing responsibility to lead the discussions and demonstrated their ability to do so, this may be taken as evidence of increased awareness of information structure. And, of course, the ITAs’ successive teaching presentations based on the texts (steps 1, 2, and 3 of Levis et al.'s Reference Levis, Muller Levis, Slater and Gorsuch2012 basic task) could be analyzed for changes or growth in their working grammatical and lexical repertoire, which would support better conceptual clarity of their presentations.

The point here would be to devise a course of treatments with the understanding that ITAs’ L2 development will take time and in-class attention. It would not be enough to simply give ITAs a one-off lecture on proper information structure in academic talks, or even giving ITAs occasional feedback on their grammar use in teaching simulations. Tracking changes in ITAs’ awareness of information structure and changes in the breadth of their working grammatical and lexical repertoire would show whether a given instructional approach, based on text-based teaching simulation preparation tasks, was effective. In a larger sense, results of such research might demonstrate that information structures, and the grammatical and lexical repertoire required to convey them, might serve as an organizing force for L2 for professional purposes speaking curricula. Typically, information structure has been seen as an organizing force in L2 writing curricula.

3.1 Tasks on the role of output in L2 language learning

ITA educators have long believed that for ITAs to improve, they need to talk. Many ITA programs use some form of simulated teaching tasks with explicit feedback as a main instructional strategy as in this example: ‘they are not just recording the audio journal, after two minutes they are going to go back and listen to it [then] every time they meet the instructor in conference [they discuss] what are they doing to meet their goals’ (Gorsuch Reference Gorsuch and Gorsuch2012a: 465). Some programs use, or have experimented with, teaching practica where low-advanced ITA candidates actually have instructional responsibilities. They are observed periodically during the semester by an ITA educator and then given feedback using video, audio, or observation notes taken during the observation (Tapper & Kidder Reference Tapper, Kidder, Kaufman and Brownworth2006). While ITA practica have been used to explore intercultural miscommunication (e.g., Tyler Reference Tyler1995), development of ITA pragmatics (Liu & Wang Reference Liu, Wang and Gorsuch2012), and ITA teacher identity formation (LoCastro & Tapper Reference LoCastro and Tapper2006), ITA practica could be a significant setting in which to examine Swain's (Reference Swain and Hinkel2005) Output Hypothesis in a non-experimental, naturalistic setting.

Swain proposes the hypothesis as ‘part of the process of second language learning’ in which learners engage in ‘processes. . .quite different than those involved in comprehending language’ (2005: 471) where learners move ‘from a primarily semantic use of language (as takes place in comprehension) to a syntactic use’ (Gass Reference Gass, Doughty and Long2003: 227). The basic premise of the hypothesis is that in interacting with other L2 speakers, L2 learners would, in some circumstances, respond to their interlocutors’ apparent non-comprehension using a modification of their original utterance which was more precise, coherent, and appropriate (Swain Reference Swain and Hinkel2005). This was taken to be evidence of an L2 speaker noticing a gap between what they wanted to say and what they actually said, which is thought to trigger processes such as learners ‘directing their attention to relevant input’ (Swain Reference Swain and Hinkel2005: 474), or testing their hypotheses about the L2, or pushing them into useful reflection on the L2 (Shehadeh Reference Shehadeh2002).

Gorsuch (Reference Gorsuch2005) did a case study with a Chinese biology ITA who was engaged in a team-teaching practicum with a native speaker TA in an undergraduate lab. The ITA had to introduce lab topics and work with students one-on-one while an experiment was underway. Through a lapel microphone, all of the ITA's utterances were recorded, and when compared to concurrent teaching observation notes, Gorsuch found evidence that the ITA engaged in self-vocalization and self-repair while lecturing on content, and during interactions with undergraduates. She argued for a broader application of the Output Hypothesis beyond learners’ immediate responses to interlocutors’ non-comprehension ‘in that the act of teaching might. . .push ITAs into communicating their ideas with greater than ordinary care in order to be comprehensible’ (Gorsuch Reference Gorsuch2005: 100). Certainly, as currently described by Swain (Reference Swain and Hinkel2005), the Output Hypothesis could apply to monologic, somewhat interactive teaching talk, and this sets the stage for a fourth task.

Researchers interested in the Output Hypothesis are not lacking research designs or data collection methods (e.g., Ellis Reference Ellis2012), any of which could be pursued within the basic premise of this task, which is to arrange teaching practica with cooperating departments, and observe and record the ITA candidates as they go about their work. These could include post-observation think-alouds, or stimulated recall sessions. For a 14-week semester, an observation session done once every two weeks would amount to seven lengthy observations and matching audio recordings and transcriptions. I think the observations are necessary to establish in which contexts the ITAs used modified speech. Was the ITA responding to non-comprehension of students either while teaching or during one-on-one interactions? Were there any obvious signals of non-comprehension to begin with? For instance, was the ITA engaging in rehearsal prior to an utterance, as the ITA in Gorsuch (Reference Gorsuch2005) apparently did while writing items on the blackboard during a small break in her pre-lab lecture? Such evidence might argue for modified output as being evidence of an internal process for L2 learning which does not necessarily require direct non-comprehension from an interlocutor, such as might be seen in experimentally prepared tasks designed to elicit modified output (e.g., Shehadeh Reference Shehadeh2002).

What if, for example, the proposed task included a permutation where ITAs doing practica knew they would engage in pre-observation planning sessions or post-observation debriefing sessions, which had the purpose of discussing two or three salient and recurring linguistic features present in the talk (see Ortega Reference Ortega and Ellis2005 for commentary on learner planning)? Might this, along with the greater pressure to communicate well as a teacher, be enough to result in modified output, and the processes underlying it, in a manner that is productive? Such pre- or post-observation sessions may provide the researcher with an array of linguistic forms to monitor for evidence of change or learning, which is still somewhat lacking in research specifically focused on the Output Hypothesis (Shehadeh Reference Shehadeh2002). This task would benefit other researchers interested in the Output Hypothesis in that additional evidence may be found for a contribution of modified output to learning over time. They may also be able to further broaden their description of the conditions under which learners are likely to engage in modified output outside of an experimental setting, which creates a useful emphasis on learner agency.

Five participants should be included in the case study in order to create perhaps more generalizable sets of data. Yet the group is small enough that a teacher/researcher, who has multiple demands on his or her time, can handle the data resulting from the observations in a somewhat timely manner, for the purposes of post-observation debriefing, for example. And there will be a lot of data.

3.2 Deliberate practice and how ITAs use learning materials

Program theory refers to how program designers, or other program participants believe their courses, or other systematic support, are supposed to help clients or learners to meet desired outcomes (Rogers Reference Rogers, Stufflebeam, Madaus and Kellaghan2000). For instance, one prevailing program theory in ITA education is in Gorsuch (Reference Gorsuch and Gorsuch2012a): three hours of language, teaching, and culture classes per week for up to two semesters should improve ITAs’ communication ability and thus result in an approval to teach. Courses should provide opportunities for controlled and free language practice which are specific to individual ITA's present problems with L2 talk (Gorsuch Reference Gorsuch and Gorsuch2012a). Thus, even though practice materials are continually sought and/or developed by ITA educators, traditional textbooks with an internal syllabus are not valued. They are seen as not sufficiently specific to individual learners’ needs.

Sustaining ITAs’ motivation is also a strong concern for ITA educators, and this plays out in prevailing program theories in a number of ways. First, there is an assumption held by some that if ITA program classes meet for more than three hours per week, ITAs will not attend (Gorsuch Reference Gorsuch and Gorsuch2012a). This may be seen as one pressure causing ITA program designers to limit time scheduled to be spent in-class. Second, ITA educators try to ensure that ITA course materials and instruction seem relevant to ITAs and their future communication needs (Gorsuch Reference Gorsuch and Gorsuch2012a). For some time, however, I have been thinking that the prevailing program theory for long-semester ITA courses may lack sufficient intensity to bring about changes in ITAs’ L2 knowledge and communication skills (e.g., Griffee et al. Reference Griffee, Gorsuch, Britton, Clardy and Ollington2008), and that ITA educators, perhaps realizing this on some level, have produced individualized materials for ITAs to use outside of class.

Instructors at several programs randomly profiled by Gorsuch (Reference Gorsuch and Gorsuch2012a) reported using a variety of online course platforms which enabled the ITAs to upload recordings of their voices engaged in some kind of practice activity. Instructors would then: (1) Listen to the audio files; (2) give feedback on the audio files, perhaps recording their own voices modeling specific features; and (3) return the feedback and recordings electronically, or face-to-face, to individual ITAs for further practice. One ITA program director saw this as positive in that increasing the amount of practice done by ITAs outside of class ought to contribute to their ability to self-monitor their language use, a skill believed to be needed for life-long learning. The energy ITA educators say they put into materials development notwithstanding, there remains the question of how ITAs actually use the learning materials at hand, and whether, in their own minds, the materials have a place in their own goals for improvement. Furthermore, if feedback accompanies the use of the materials, is it enough, and of the kind the ITA feels is useful? Further, do ITAs form their own goals, and if so, are they aligned to actual potential improvements to their talk or teaching? If ITAs are supposed to work on their own for significant amounts of time due to prevailing ITA program theories, these are salient questions, and ones which ITA educators do not necessarily pose (Gorsuch Reference Gorsuch and Gorsuch2012a). L2 educators and materials writers do not necessarily evaluate the effectiveness of their materials (Bautista Reference Bautista, Hildago, Hall and Jacobs1995), and are likely unaware of how learners are using the materials they assign or design (Gorsuch Reference Gorsuch and Gorsuch2012a). And, while ITA educators think offering feedback is a key element of their teaching, there is probably little tracking of the frequency and type of feedback ITAs get, and how ITAs might use the feedback (Gorsuch Reference Gorsuch and Gorsuch2012a).

Enter deliberate practice theory, which posits that expert performance for medical doctors, musicians, insurance sales personnel, chess players, and other adult learners is attained through defining specific goals to improve performance, and engaging in repetitive ‘practice tasks that allow them to refine their knowledge and skills’ (van de Wiel et al. Reference van de Wiel, Van den Bossche, Jannsen and Jossberger2011: 82). Expertise is not innate, but must be cultivated (Ericsson Reference Ericsson2004). The tasks themselves must be expressly for ‘the aim of learning or improving professional competence’ and include ‘preparation, mental stimulation, asking for feedback and advice, evaluation, reflection and updating activities’ (van de Wiel et al. Reference van de Wiel, Van den Bossche, Jannsen and Jossberger2011: 83). Learners cannot achieve expertise simply by reacting to events in a workplace, but rather by refining personal understandings of what they need to improve, often with the help of teachers or coaches, and then engaging consciously in repetitive, self-assigned tasks to improve in those self-selected areas (van de Wiel et al. Reference van de Wiel, Van den Bossche, Koopmans, Dochy, Gijbels, Segers and den Bossche2012). This suggests growth in learners’ ability to self-regulate, and self-generate ‘thoughts, feelings, and actions that are strategically planned and adapted to the attainment of personal goals’ (Zimmerman Reference Zimmerman, Ericsson, Charness, Feltovich and Hoffman2006: 705). As Ericsson notes (Reference Ericsson2004: S 73): ‘Engaging in practice activities with the primary goal of improving some aspect of performance is an integral part of deliberate practice.’ This sets the context for the next research task.

Like most longitudinal projects, this overarching research task can be broken into sub-tasks, as suggested above. The following procedures address the first and second sub-tasks, and will generate data to create a context for the third sub-task. Working over a period of at least one year, researchers need to follow a group of learners, collecting data using learner diaries, questionnaires, and interviews. In particular, interview protocols need to directly refer to whatever language practice materials ITAs say they are using. Such interviews may include a think-aloud in which an ITA candidate shows the materials he or she uses, and talks through how he or she uses it. This may provide needed detail on what feedback they can reliably get on performances resulting from their use of the materials, and how they use the feedback. ITA candidates should be queried on how they developed their practice methods, and how they believe their practice method helps them. For the third sub-task, ITAs’ formation of goals concerning their professional communication needs can be tracked, again using learner diaries, questionnaires, and interviews. These data collection instruments can be designed to address specific questions, for instance, do ITAs’ communication goals become more specific over time? In what ways? Do the general areas of their concerns change? In response to what? Do their choices of practice materials and manner of use change in response to goals? Are the materials they use relevant to their goals? Are the ways in which they are asked to use materials relevant to their goals? Is there any evidence, perhaps gathered by video or audio recording ITA candidates’ teaching or teaching simulations over time, that language features ITAs self-select for practice change or improve?

Research of this type would reveal a few things relevant to ITA educators. ITAs may be using assigned practice materials, but in ways educators do not expect. ITAs may also be using self-selected materials according to more or less well-defined goals ITAs have formed for themselves. ITA educators may find value in ITAs’ choices of materials, and their goals. Certainly, findings from this research would, I hope, demonstrate the importance of keeping track of emerging or persistent issues in ITA performance in talk and teaching arising in feedback sessions. ITAs may learn to keep track of these issues for their own purposes. Sadly, it might also be found that whatever materials ITAs use, and how often, and in what manner, it is not enough to compensate for time spent out of class due to prevailing ITA program theories. Finally, the findings may lead ITA educators or researchers to more thoroughly explore self-regulation among ITAs, both during and after ITAs’ participation in preparation or in-service courses.

Researchers working with deliberate practice theory have their own issues to ponder, and research with ITAs as participants may inform their research. Currently their research does not include populations who need to develop their L2s for the purposes of teaching. It may be that deliberate practice is effective for only a limited number of agreed-upon attributes of expertise, whereas ITAs may have whole constellations of areas in their performance they need to improve. Further, publications on deliberate practice do not necessarily describe how the attributes of expertise being sought are defined, and by whom. For instance, do definitions of expertise require collaboration with learners in order to be effective as desired targets for improvement? Research done with ITA educators involved, and with ITAs as participants, may reveal how expertise is most usefully defined. Finally, the literature on deliberate practice does not reveal in detail whether or how this theory can be applied to large programs. Doing this research might demonstrate whether ITAs, or teachers and coaches, can sustain needed levels of focused feedback working with groups of ten or more learners.