In a recent study I used audio-supported repeated reading (RR) and awareness-raising tasks for a 14-week semester to help ITAs improve their discourse intonation (DI). I believe that DI is difficult to learn as implicit knowledge. Further, any changes in DI and spoken fluency might be ITAs having greater or less success applying explicit knowledge to their talk.

New Numbers and Known Things
In 2011–2012, there were 300,430 international graduate students enrolled in U.S. universities, and 164,394 were supported by those schools, likely as teaching assistants (Institute of International Education, 2013). Ninety percent came from countries in which English is not widely used (Institute of International Education, 2013). At many institutions, many undergraduate science and math courses are taught by ITAs. Indeed, ITAs make undergraduate education possible and as instructors are contributors to undergraduate learning.

ITAs must teach using English, their L2. Although many ITAs have developed some control over L2 lexis and syntax and are high-intermediate to advanced learners, they have little experience using their spoken English for social and instructional purposes. Thus ITAs struggle with spoken fluency, in particular prosody.

DI and ITAs
DI is the use of pause groups (associated with fluency) and sentence-level stress and tone choices (associated with prosody) for communicative purposes. DI is used to emphasize and differentiate ideas, begin and end topics, and express social relationships (Pickering, 1999). Evidence suggests that the ability to use DI is essential for ITAs’ success as instructors (Hahn, 2004), yet DI is acquired late and with difficulty (Pennington & Ellis, 2000). Because use of appropriate DI is a basis for listener perceptions of spoken ability, we need to devise and test pedagogical interventions to bring about improvements in learners’ DI.

ITAs’ Implicit and Explicit Knowledge
A few years ago I came to the conclusion that I needed to work with ITAs’ implicit and explicit knowledge. Prosody (DI) is an abstract system and is hard for learners to apply to language use. If ITAs have only explicit knowledge (what we teach and have them practice), they cannot apply what they know while talking due to L2 processing constraints. So if they are struggling with what they want to say in the L2, anything we teach them about using prominence on important words, or not breaking up idea units with pauses, will go out the window. The “default processing mode” for L2 acquisition processes “is implicit” (Doughty, 2003, p. 292). To use DI requires complex knowledge, which according to Doughty (2003) is acquired as “implicit knowledge leading directly to procedural ability” (p. 291). At the same time, DI should also be taught in order to raise learners’ awareness. This would result in explicit, declarative knowledge. Over time and with practice, this knowledge may become proceduralized and available for talk.

The Study
In 27 treatments, seven ITAs in an experimental group did combined implicit knowledge–building awareness-raising tasks and audio-assisted RR treatments. They also had 14 weeks of explicit DI instruction. They were L1 Mandarin speakers and had been in the United States an average of 8.43 months. They had not “passed” an intensive workshop and semester-long course, and in the parlance of the ITA program had “stalled out” with SPEAK test scores averaging 40. The SPEAK test (ETS, 1996) is a spoken performance test, and a score of 40 means an ITA lacks the communicative competence to explain content and manage classrooms.

I worked with the experimental group’s explicit knowledge by using materials focused on thought groups, prominence, and tone choices that linked specific forms to meaning. I used listening tasks based on authentic classroom recordings. ITAs engaged in rehearsed and free speaking practice, teaching simulations, and feedback sessions.

I worked with the experimental group’s implicit knowledge using awareness-raising tasks and RR sessions. Each awareness-raising task included a debriefing session in which participants were asked to speculate on how the feature was related to the speaker’s intended meaning.

Then came audio-supported RR treatments:

1. Participants silently read a segment of a 500-word popular science text once. Instructors answered any questions on word meaning and other issues.
2. Participants read the same text a second and third time while listening to an audio file of the text.
3. Participants silently read the text a fourth time.

The texts were 500-word segments of science texts taken from Science for Students. These are original pieces on astronomy, chemistry, and other subjects at the secondary school level. They needed to be easy enough so that the input from them would be comprehensible and learners could focus on specific DI features in the audio model as they built up experience with the text through the repetitions.

For comparison, seven ITAs who were newly arrived in the United States formed a control group. They also had SPEAK scores of 40 and were L1 Mandarin speakers. The control group had 3 weeks of explicit DI instruction, but no input treatments.

I wanted to know if there were changes in both groups’ use of DI in audio-recorded and transcribed parallel read-aloud and free-response tasks. The read-aloud task would capture changes in ITAs’ explicit knowledge, and the free-response task would capture changes in ITAs’ implicit knowledge. For both groups the pre- and posttest read-alouds were paragraphs from freshman texts in their disciplines. Participants read their passage aloud for both the pre- and posttest. The free-response task was participants’ recorded extemporaneous response to the query “Tell me about your research.”

The participants’ talk was transcribed. The audio files and transcripts were randomly assigned a code number so that I did not know whether I was analyzing the talk of an experimental or control group member, or of a pre- or posttest. The audio files and transcripts were analyzed for speech rate, pause groups, prominence, and tone choices.

Results
The good news was that explicit and implicit instruction in DI did not slow down participants’ speech rates from the pretest to the posttest on both read-aloud and free-response tasks, suggesting that learners did not divert attentional resources to a focus on form (DI). In the read-aloud condition in which processing burdens were reduced, both groups improved slightly over time on planning versus hesitation pauses, prominence, tone choices, and length of tone choice groups. This suggests participants had sufficient attentional resources, and explicit knowledge of DI, to apply DI to their talk.

In the free-response task, processing burdens were increased as participants in both groups had to encode their own speech. On most measures, there was little evidence of change in implicit knowledge of DI for either group. Both groups sounded choppy and largely monotone. However, the experimental group’s speech was more “musical,” with a greater variety of rising and falling tones, and they encoded longer rising tone pause groups, suggesting they had some implicit knowledge of the discoursal value of rising tones. These improvements were slight, however. Clearly, implicit knowledge is difficult to form.

Suggestions for Instruction
We should continue instruction in DI. In some conditions ITAs are able to use explicit DI knowledge. Implicit knowledge growth is slow, but this is not a reason not to find ways to develop learners’ implicit DI knowledge. This is especially true if done in pedagogically worthwhile ways. The RR treatments are pedagogically worthwhile, because they support vocabulary building (so ITAs know where to use prominence) and provide consistent models of DI connected to discoursal meaning. Perhaps the RR texts should be limited to 300 words; the experimental group may have had too many things in the visual and aural input to pay attention to.

What This Means for ITA Programs
As a colleague put it, “acquiring L2 implicit knowledge is really just a hard, slow slog through a muddy field.” Many aspects of L2 development are slow, and this is a fact that cannot be finessed. There is no magic bullet. Further, our classes are not intensive enough to bring about improvements in ITAs’ DI. Many ITA programs expect their ITAs to practice outside classes, but I do not know whether ITAs do practice, nor whether the practice is useful. Faced with the slowness of ITAs’ L2 development (see TESOL’s Position Statement on the Acquisition of Academic Proficiency in English at the Postsecondary Level), ITA programs need to have the best theory-driven answers with which they can defend themselves and explain to impatient administrators why ITA courses are an essential part of any “internationalization” plans universities hold dear. We need to characterize L2 learning as something different than learning content such as history, and as proceeding more slowly with greater in-class and out-of-class effort. To administrators, it may seem that our ITA programs are ineffective, when it is in fact the slowness of L2 learning which is salient, and also sometimes the fact that some ITAs come to us with low abilities.

References
Doughty, C. (2003). Instructed SLA: Constraints, compensation, and enhancement. In C. Doughty & M. Long (Eds.), The handbook of second language acquisition (pp. 256–310). Malden, MA: Blackwell.

ETS. (2009). TSE—Test of Spoken English. Retrieved from http://www.ets.org/portal/site/ets/menuitem.fab2360b1645a1de9b3a0779f1751509/?vgnextoid=b5d7d898c84f4010VgnVCM10000022f95190RCRD

Hahn, L. (2004). Primary stress and intelligibility: Research to motivate the teaching of suprasegmentals. TESOL Quarterly, 38, 201223.

Institute of International Education. (2013). Open doors 2011/2012 fast facts. Retrieved from http://www.iie.org/Research-and-Publications/Open-Doors

Pennington, M., & Ellis, N. (2000). Cantonese speakers' memory for English sentences with prosodic cues. Modern Language Journal, 84, 372–389.

Pickering, L. (1999). An analysis of prosodic systems in the classroom discourse of native speaker and nonnative speaker teaching assistants (Unpublished doctoral dissertation). University of Florida, Gainesville, FL.