Language Learning
ISSN 0023-8333
Teacher- and Learner-Led Discourse in Task-Based Grammar Instruction: Providing Procedural Assistance for Morphosyntactic Development
Paul D. Toth
Temple University
For many years, task-based second-language (L2) grammar instruction has been considered the ideal means for achieving a focus on linguistic form within meaningful, purposeful communication (Ellis, 2003; Long & Crookes, 1993; Nunan, 1989; Samuda & Bygate, 2008).
Within this framework, small-group, learner-led discourse (LLD) is often believed to facilitate L2 development better than whole-class, teacher-led discourse (TLD), given the greater discursive autonomy for learners and presumed greater opportunities for negotiated interaction (Lee, 2000; Long & Porter, 1985; Pica, 1987; van Lier, 1996).
Indeed, many have criticized TLD, given evidence of disfluent exchanges during which teacher turns impede rather than support learner participation (e.g., Brooks, 1993; Donato & Brooks, 2004; Hall, 1995, 2004; Leemann-Guthrie, 1984).
Still, some have argued that TLD may greatly benefit learners if teacher assistance, negotiation, and feedback are well managed (Adair-Hauck & Donato, 1994; Ant´ n, 1999; McCormick o & Donato, 2000) and that witnesses to such exchanges may benefit as much as active participants (Ohta, 2001).
However, only a few studies have directly compared TLD with LLD, and these have yielded conflicting results (Fotos, 1993, 1994; Pica & Doughty,
I am greatly indebted to the Spanish students at the University of Pittsburgh and the University of Akron who participated in this study, as well as the instructors who graciously volunteered much time and effort to implement the treatment conditions, including Joyce Rowan, Fernando Feliu, Jessie Carduner, and Tim Aiken. A number of individuals also gave helpful advice, comments, and feedback on the design of this study since its inception as a dissertation project, as well as in its current format as a journal article. Special thanks are due to Alan Juffs, Richard Donato, Merrill Swain, Teresa Pica, and three anonymous reviewers. Invaluable statistical advice was provided by Carol Baker at the University of Pittsburgh, Richard Steiner at the University of Akron, and Amy Atwood and Edward Taylor at the University of Wisconsin-Madison. Any remaining errors are my own. Correspondence concerning this article should be addressed to Paul D. Toth, Department of Spanish and Portuguese, Temple University, Anderson Hall 4th Floor, West Berks Street, Philadelphia, PA 19122-6090. Internet: [email protected]
Language Learning 61:Suppl. 1, June 2011, pp. 141–188 C 2011 Language Learning Research Club, University of Michigan
141
Toth
Teacher- and Learner-Led Discourse
1985; Van den Branden, 1997, 2000).
This study therefore aimed to contribute to the literature by presenting quantitative and qualitative classroom data gathered under similar task conditions for both TLD and LLD. Rather than unequivocally advocating for one discourse format over the other, I evaluated the strengths and limitations of both as tools for L2 morphosyntactic development. Participants in the study included 78 English-speaking adults from six university classes of beginning L2 Spanish, with two assigned to each treatment (LLD = 25; TLD = 28) and two other classes comprising a control group (n = 25).
Instruction involved seven lessons targeting the anticausative clitic se, with one recorded and transcribed in each treatment. Results on grammaticality judgment and guided production tasks administered as a pretest, posttest, and delayed posttest indicated a stronger performance for TLD learners on both tasks. Although the transcript data revealed numerous episodes of constructive L2 negotiation and support in LLD—often addressing topics that were not assessed quantitatively—the transcripts nonetheless suggest that the teachers facilitated L2 development by directing attention more consistently to target structures and providing morphosyntactic “procedural assistance” to learners during utterance formulation. Because such moves are unlikely to occur among learner peers alone, I argued that teachers may be uniquely positioned to assist L2 development through their discursive role, such that an ideal task-based L2 pedagogy would include principled sequences of both TLD and LLD. Keywords classroom research; teacher-led discourse; learner discourse; task-based instruction; instructed SLA; L2 Spanish; anticausativity; pushed output; proceduralization; scaffolding
Much recent research on second language (L2) instruction has suggested that drawing learners’ attention to connections between formal properties of the L2 and the meanings they convey facilitates L2 morphosyntactic development (Doughty & Williams, 1998b; VanPatten, 2004a).
Within the interactionist model of acquisition, this is said to occur when communication difficulties between speakers lead to conversational modifications that enhance the comprehensibility of both the language that learners hear and that which they produce (Long, 1981, 1996).
This negotiation of meaning is said to simultaneously cause learners to attend to, or focus on, L2 form as a means of accomplishing their desired communicative act (Doughty, 2001).
Thus, where classroom instruction is concerned, meaningful interaction is held to be an essential ingredient for L2 development, especially when it entails modified, comprehensible input, opportunities for purposeful output, and feedback on formal accuracy (Ellis, 1997; Gass, 2003).
Although the utility of implementing L2 grammar instruction in this way, as a “focus on form” (Doughty & Williams, 1998a; Long, 1991), has been documented both in the context of teacher-led, whole-class discourse as well as small-group interactions among learners (e.g., Ellis, 2003;
Language Learning 61:Suppl. 1, June 2011, pp. 141–188 142
Toth
Teacher- and Learner-Led Discourse
Norris & Ortega, 2000; Ohta, 2001), little work has made direct comparisons of learning outcomes for these two primary formats within which L2 grammar instruction most often takes place. This article therefore aims to contribute to the literature by considering both quantitative and qualitative data from teacher- and learner-led discourse (henceforth, TLD and LLD, respectively) during a week of beginning-level, university L2 Spanish instruction on the pronoun se. Task-Based Instruction and Classroom Discourse Formats For a number of years, L2 educators have advocated a task-based approach to grammar instruction as the ideal way to accomplish a focus on form within meaningful, purposeful communication (Breen & Candlin, 1980; Ellis, 2003; Lee, 2000; Long & Crookes, 1993; Nunan, 1989; Skehan, 1996; Van den Branden, 2006).
Based in part on a view of natural communication as a series of intentioned “speech acts” (Austin, 1962; Hymes, 1972; Searle, 1969), task-based lessons are organized into discrete acts of communication such as identifying, describing, comparing, evaluating, summarizing, or narrating, often in response to images, texts, or realia from native-speaking target-language communities. Tasks might also include content designed to heighten sensitivity to cultural perspectives within the L2 speech community and/or engage learners in problem-solving related to other academic subjects (National Standards in Foreign Language Learning Project, 1999; Pica, 2005).
Once a task has been chosen, teachers spend class time equipping learners for performance by highlighting necessary L2 structures and then having learners interact with varying amounts of support and feedback, ideally leading to an autonomous performance. Often, support is provided by breaking the task down into simpler subcomponent steps, with trial runs and feedback given before independent interaction, as well as materials containing relevant linguistic, contextual, and procedural cues (Samuda, 2001; Skehan, 1996, 1998).
Whatever the task format or objective, however, the essential role of L2 grammar within this approach is to support task performance rather than to serve in and of itself as the organizing feature for classroom discourse (Long, 1991; Long & Crookes).
Despite a general agreement on the role of grammar instruction in taskbased teaching, much recent work on L2 grammar instruction has been more focused on the relationship between learning outcomes and specific interactional features—such as the quantity and quality of positive or negative evidence—than with the contribution of the discourse format to these outcomes (cf. Leeman, 2003; Lyster, 2001; Mackey, 1999).
In one example of
143 Language Learning 61:Suppl. 1, June 2011, pp. 141–188
Toth
Teacher- and Learner-Led Discourse
earlier work that did make such comparisons, Pica and Doughty (Doughty & Pica, 1986; Pica, 1987; Pica & Doughty, 1985) investigated the quantity and quality of discourse features claimed to benefit L2 acquisition, but did not assess learning outcomes. All three studies were short-term investigations of intermediate L2 English classrooms in which similar tasks were performed first as TLD and then as LLD. In all three, they found that although LLD consistently provided learners with more speaking turns, beneficial conversational adjustments could be found to varying degrees in either format, depending on the task design. For example, Pica (1987) reported that when the task involved open-ended, collaborative decision-making, LLD yielded somewhat fewer adjustments (6% of learner utterances vs. TLD’s 11%), given that learners in small groups could easily opt out of participation. However, when the task required an information exchange, LLD yielded many more modifications (24% of learner utterances versus vs. TLD’s 15%), because each learner was required to contribute. Doughty and Pica then showed that negotiation in LLD increased further when learners were arranged in dyads rather than in small groups.
In all three studies, the authors concluded that with adequate task design, LLD creates participant roles that more aptly allow for negotiation than TLD. However, Pica (1994) later argued that documenting such theoretically justified modifications ought to be followed by independent measures of learning outcomes in order to affirm that negotiation does in fact lead to morphosyntactic development. Although many recent studies have indeed provided such evidence in LLD contexts (see Ellis, 2003, for a review), few have compared results for similar tasks performed as TLD. Two important exceptions to this lack of comparison work are Van den Branden’s (1997, 2000) research in multilingual upper elementary classrooms in Belgium and Fotos’s (1993, 1994) work in university-level L2 English classes in Japan. Van den Branden investigated the impact of teacher versus peer negotiation on L2 Dutch children’s performance on reading comprehension (2000) and oral picture description tasks (1997).
He found in both cases that negotiation occurred most frequently and learning outcomes were strongest when L2 learners interacted with a teacher/researcher rather than with native language (L1) Dutch-speaking peers. Contrary to Pica and Doughty, Van den Branden attributed his results to a “pupil-centered” yet systematic “pushing” of learner output that could only be accomplished by an interlocutor with a pedagogical purpose (1997, p. 628).
Fotos’s studies in adult learning contexts were less conclusive. She compared TLD to LLD in grammar-oriented communicative tasks during 9 weeks of instruction, with a third group given meaning-oriented tasks as LLD with no overt grammatical focus. Using a pretest, posttest, and
Language Learning 61:Suppl. 1, June 2011, pp. 141–188 144
Toth
Teacher- and Learner-Led Discourse
delayed posttest design, she found equal gains for grammar-oriented TLD and LLD on grammaticality judgment and production tasks, with meaning-oriented LLD lagging behind. However, in the earlier study, Fotos also assessed the frequency of learners’ “noticing” of three target structures, operationalized as the ability to underline any “special uses of English” appearing in a printed narrative (1993, p. 390).
The results showed the two grammar groups (LLD and TLD) again outperforming the meaning-oriented LLD group, but no differences between them except that TLD learners noticed more tokens of one of the three target grammar items. In explaining the difference, Fotos (1993) speculated that her dual role as both researcher and TLD instructor might have influenced the outcome. Given equal gains on the other tasks, however, she argued that LLD was at least as effective as TLD in facilitating morphosyntactic development. Strengths and Limitations of LLD and TLD Despite the lack of conclusive evidence indicating superior outcomes for either LLD or TLD, some have explicitly advocated that small-group or dyadic interactions be the essential format for task-based L2 instruction. Often, it has been claimed that LLD is more essentially communicative than TLD, in that it better prepares learners for the negotiated interactions that take place outside the classroom (Jacobs, 1998; Lee, 2000; Long & Porter, 1985; Nunan, 1987, 1989).
Meanwhile, others have argued that LLD confers greater linguistic autonomy, creativity, and self-regulation upon learners, all of which are necessary if they are to master the L2 as a tool for communication (Brooks, Donato, & McGlone, 1997; Donato, 1994; van Lier, 1996).
Early proposals for task-based curricula furthermore viewed LLD as a logical response to research documenting gaps between direct instruction on L2 forms and learners’ ability to use them in obligatory contexts (cf. Breen & Candlin, 1980; Krashen, Sferlazza, Feldman, & Fathman, 1976; Larsen-Freeman, 1975; Long, 1985).
Likewise, many of the instructional benefits ascribed to the teacher talk of TLD were also documented during LLD, namely comprehensible input, modified interaction, and negotiation over L2 form (Doughty & Pica, 1986; Pica & Doughty, 1985; Pica, Holliday, Lewis, & Morgenthaler, 1989; Porter, 1986; Varonis & Gass, 1985).
Furthermore, TLD in some studies was shown to be a less-than-ideal context for developing L2 grammatical and communicative competence, given the dominance of the teacher in shaping discourse and the impossibility of engaging all students in meaningful language use (Hall, 1995, 2004; Lee, 2000; Leemann-Guthrie, 1984; Musumeci, 1996; Nunan, 1987).
145 Language Learning 61:Suppl. 1, June 2011, pp. 141–188
Toth
Teacher- and Learner-Led Discourse
Even during open-ended discussions, Lee, for example, observed that too often TLD “degenerates into a conversation between the instructor and the two best learners in the class,” effectively “disenfranchising” the other learners and allowing them to be “cognitively [un]engaged” (p. 33).
With regard to LLD as grammar instruction, however, similar concerns have arisen that learners might not all be fully engaged in the instructional goal (Ellis, 2003).
It has been observed that learners at times produce minimal L2 utterances in order to expediently accomplish the task (Seedhouse, 1999), that they sometimes provide poor L2 models for each other (Prabhu, 1987), and that they might be more inclined to focus on lexical rather than morphosyntactic issues when negotiating over form (Buckwalter, 2001; Morris, 2002; Williams, 1999).
Proponents have argued, however, that many of these concerns can be addressed by adequate task design. For example, Pica, Kanagy, and Falodun (1993) claim that because “information gap tasks” require a two-way exchange leading to only one acceptable outcome, negotiation over L2 form is more likely. Furthermore, transcripts from more open-ended, collaborative tasks requiring editing or composing in the L2 have shown that learners can indeed serve as effective assistants to peers in producing, analyzing, and correcting L2 structures (Donato, 1994; Foster & Ohta, 2005; Swain, 1998, 2000; Swain & Lapkin, 1995, 2001).
It is also claimed that specific L2 structures can be effectively targeted if they are either “useful” or “essential” to the communicative act being performed (Fotos, 2002; Loschky & Bley-Vroman, 1993)1 and that learners might employ target structures more fluently if pretask activities prime their attention to the necessary task language and posttask follow-ups hold them accountable for adequate task performance (Skehan, 1996, 1998; Skehan & Foster, 1999).
Pica, Kang, and Sauro (2006) indeed found a strong relationship between negotiations that had occurred during information gap tasks and L2 English learners’ immediate recall of connections between target forms and their meaning. Because LLD engages all learners simultaneously in interaction, it can be argued that as long as the task effectively focuses attention on L2 form-meaning relationships, the sheer mathematics of LLD’s interactional opportunities favors this format over TLD as a tool for L2 morphosyntactic development. Nonetheless, in actual practice, much L2 grammar instruction often occurs as whole-class TLD, and research to date has yielded disparate conclusions about its benefits and drawbacks. It has long been observed that TLD often yields a recurrent cycle of teacher “Initiation” moves (i.e., questions or prompts directed at the class), followed by learner “Responses,” and subsequent teacher “Feedback” (Mehan, 1979; Sinclair & Coulthard, 1975).
One disadvantage of
Language Learning 61:Suppl. 1, June 2011, pp. 141–188 146
Toth
Teacher- and Learner-Led Discourse
these “IRF” sequences, in addition to yielding fewer turns per learner than LLD, is that they are often used by teachers during grammar instruction to limit learner participation to the mere recitation of vocabulary and grammatical paradigms, with little opportunity to develop a broader interactional competence (Brooks, 1993; Hall, 1995, 2004; Nunan, 1990).
Still, others have argued that it is not the IRF pattern itself that makes TLD problematic but rather teacher turns within IRF that either fail to establish cohesion to a broader discourse purpose, or that too narrowly frame desirable learner responses (Cullen, 1998; Johnson, 1995; Toth, 2004; Wells, 1998).
When teacher utterances more consistently adhere to communicative task goals and build upon the topical content of learner utterances, research has indicated that fewer comprehension difficulties (Toth) and greater numbers of learner-initiated turns (Johnson) result. Indeed, teacher turns occurring in the context of whole-class collaborative interactions have also been shown to “scaffold” learners through tasks that they could not have performed on their own, thus deepening their engagement with needed L2 forms (Adair-Hauck & Donato, 1994; Ant´ n, 1999; McCormick & Donato, o 2000).
Ohta’s (2000, 2001) research on feedback during TLD further suggests that learners not directly participating in classroom discourse might nonetheless be cognitively engaged as the interaction unfolds. After equipping adult L2 Japanese learners with individual microphones, her recordings of their subvocal, private speech during whole-class interactions showed that many nonparticipants were actively hypothesizing about L2 form-meaning relationships while feedback was provided to others. If indeed nonparticipants might be quietly mapping out L2 form-meaning relationships during TLD, then this would mitigate its disadvantage in providing fewer turns per learner, as well as Lee’s (2000) concern that nonparticipants are effectively disenfranchised. Given Swain’s (1995, 1998, 2000) contention that L2 output benefits grammatical development by “pushing” learners to test hypotheses, notice gaps in their knowledge, and engage in metalinguistic reflection, Ohta’s (2000, 2001) work suggests that well-implemented TLD might indeed engage learners in this mental activity, even when the teacher does not directly call on them to respond. The Present Study Design and Research Questions The present study investigates the relationship between classroom discourse format and L2 morphosyntactic development by comparing Spanish classes
147 Language Learning 61:Suppl. 1, June 2011, pp. 141–188
Toth
Teacher- and Learner-Led Discourse
where similar tasks were performed as either small-group LLD or wholeclass TLD. The data to be considered here were gathered as part of a larger study (Toth, 1997) that investigated not only improvements with the object of instruction but also changes in L1 transfer and overgeneralization errors.2 Although postinstruction improvements with se revealed differences between the treatment groups, significant decreases in L1 transfer and increases in overgeneralization were statistically similar. Thus, because group contrasts appeared to be related more to the way learners responded to object of instruction itself, rather than to its concomitant repercussions in inducing or expunging other grammatical errors, this study will, due to space limitations, focus only on the results of the larger study that pertain to improvements with se. A full reporting of the learners’ overgeneralization and transfer errors can be found in Toth (1997, 2000, 2003).
Given that many of the strengths and weaknesses identified for either LLD or TLD have as much to do with adequate instructional design as inherent discourse features, implementation of LLD and TLD will necessarily require that each treatment reflect current proposals for maximizing meaningful language use and attention to L2 form, in order to avoid confounding factors that would render either one a straw man. Thus, following the work of Pica and her colleagues (Pica et al., 1993, 2006), LLD will be implemented strictly as two-way information exchanges, with Spanish se being either “useful” or “essential” to the task (Fotos, 2002; Loschky & Bley-Vroman, 1993).
Furthermore, task performance will be preceded by a pretask warm-up for the planning and priming of necessary language, followed by a posttask follow-up to lend accountability to its completion (Skehan 1996, 1998).
Meanwhile, TLD will be designed to avoid the noncohesive “local lexical chaining” of teacher questions observed by Hall (1995, p. 44) during attempts to elicit target forms. Instead, TLD instructors will use task topics and goals as the basis for building cohesion, with multiple responses elicited for each teacher initiation and scaffolded feedback given to facilitate form-meaning mapping, as in Adair-Hauck and Donato (1994) and Ant´ n (1999).
Although the task goal in each group o will be identical (i.e., describing a picture, narrating an event, comparing two halves of an information pool), the participatory structure of TLD will necessarily involve more open-ended, collaborative turn-taking than LLD, given the difficulty of designing numerous whole-class information exchanges for 20 or more learners.3 Thus, whereas the information gap tasks of LLD will attempt to regulate turn-taking and attention to L2 form-meaning connections via task design, responsibility for such regulation under the more collaborative structure of TLD will necessarily rest with the teacher. Therefore, a teacher’s
Language Learning 61:Suppl. 1, June 2011, pp. 141–188 148
Toth
Teacher- and Learner-Led Discourse
skill in discourse management will have to be more effective than that of the information-gap design in order to offset the greater number of learner turns and greater learner control over discourse that LLD entails. The dependent variable in this study—development in knowledge of Spanish anticausative se—will be operationalized as performance on a timed, written test including a grammaticality judgment (GJ) task and a sentence-level picture description task. Although the validity of any GJ task ultimately hinges on the particulars of its design and implementation, the present task’s validity in assessing L2 grammatical development can be inferred from the larger related study (Toth, 1997, 2000, 2003), in which group results for the GJ task revealed overgeneralizations of se that could only be explained by underlying linguistic knowledge, rather than explicit instruction, L1 transfer, or L2 input data. Thus, it is unlikely that learners will have simply answered blindly or relied exclusively on remembered explicit information for their answers. Hence, results for improvements with se on the GJ task will be assumed to represent a performance that draws at least in part on internalized L2 grammatical knowledge. The GJ task will also triangulate the production task with one that does not as closely reflect the instructional practices of the treatment groups. Furthermore, qualitative data will be provided to shed light on the interaction patterns in each group during a picture description activity implemented as whole-class interaction in TLD and as an information gap in LLD. However, note that because the study involves four course sections meeting for 7 total hours each, it will not be possible to quantify all tokens of se to which learners are exposed or the quantity of interaction in which each participant engages. Thus, the limited qualitative data will be used here simply to characterize the support and feedback provided to learners during task performance under either condition. The research questions are thus as follows: Question 1: Will LLD provide an advantage in grammaticality judgments for Spanish anticausative se when compared to TLD? Question 2: Will LLD provide an advantage over TLD in performance with anticausative se on sentence-level picture descriptions? Question 3: Will excerpts of classroom interactions reveal differences in the way learners in each group attend to the form-meaning relationships associated with anticausative se and use the target form for output? Method Participants Participants in the two experimental treatments included 53 native Englishspeaking adult learners in four intact classes of a second-semester, beginning
149 Language Learning 61:Suppl. 1, June 2011, pp. 141–188
Toth
Teacher- and Learner-Led Discourse
L2 Spanish course in a large, public university in the United States. The LLD and TLD groups each consisted of two classes taught by different instructors in order to control for variation in teaching style and class rapport (LLD: n = 25; TLD: n = 28).
An additional 25 English-speaking adult learners from two classes of second-semester Spanish formed a control group (C) at a second large, public American university nearby, in which the L2 Spanish curriculum was nearly identical to that of the first. Most learners in the three groups had taken some Spanish in secondary school (LLD: n = 20; TLD: n = 25; C: n = 19), with the overwhelming majority having studied for 1 or 2 years. However, prior departmental placement testing at both universities had determined that none were proficient enough to take more advanced L2 Spanish courses, and participant questionnaires revealed that none had been exposed to Spanish at home. Nonetheless, to ensure the absence of false beginners prior to instruction, any participant who scored 50% or higher for anticausative se on the picture description portion of the pretest was eliminated from the study. Hence, the participant numbers reported here reflect the elimination of seven participants for this reason, along with a number of others who were excluded due to absenteeism. Finally, a native-speaker comparison group comprised of 30 adult Spanish native speakers from all parts of Latin America also participated. A questionnaire showed that none of them had any significant experience teaching Spanish, and so it was assumed that any awareness of pedagogical rules for se would not affect their test performance. Spanish Anticausative se Within generative syntactic theory, most linguists have argued that a primary function of Spanish se is to derive variations in sentence meaning by “absorbing” one of the noun phrase (NP) arguments associated with a given verb (Dobrobie-Sorin, 1998; Montrul, 2004; Raposo & Uriagereka, 1996).
This derivational process can be seen in Examples 1a–1c by comparing the transitive sentences on the left side of each arrow with their “detransitivized” counterparts on the right. Thus, in Example 1a, when the patient of the sentence on the left is absorbed, leaving only an agent plus se on the right, a reflexive or reciprocal meaning is derived. In Example 1b, meanwhile, the same sentence on the left yields a passive or impersonal reading when the agent is absorbed and se appears on the right with the patient as postverbal subject. Finally, absorption of the agent with a different verb in Example 1c derives the possibility of a middle voice reading, in addition to the passive or impersonal, given that the meaning of cocinar “to cook” implies the possibility of a spontaneous, inchoative event when used intransitively (Levin & Rappaport-Hovav, 1995; Pustejovsky, 1995).
Language Learning 61:Suppl. 1, June 2011, pp. 141–188 150
Toth
Teacher- and Learner-Led Discourse
Thus, by detransitivizing verbs in this way, se derives a number of possible meanings in Spanish that depend on the subtle semantic properties of verbs and the thematic roles they assign to NPs. For English-speaking learners, acquisition is further complicated by the fact that corresponding morphology in the L1 varies considerably. Nonetheless, the distinctions made in many instructional materials among a passive, impersonal, and middle voice se use English-based classifications that belie a unitary Spanish phenomenon embracing both inchoative and externally caused events. Throughout this study, then, although the learners’ curricular materials might make reference to a passive, impersonal, and middle voice se, the term “anticausative se” will refer collectively to all of these cases where se absorbs the agents of verbs and patients become subjects. Example 1 a. Ellos prepararon la comida. AGENT PATIENT “They prepared the food.” Ellos se prepararon. AGENT “They prepared themselves/each other.” b. Ellos prepararon la comida. → Se prepar´ la comida. o AGENT PATIENT PATIENT “They prepared the food.” “The food was prepared/One prepared food.” c. Ellos cocinaron la comida. → Se cocin´ la comida. o AGENT PATIENT PATIENT “They cooked the food.” “The food Ø cooked/was cooked/ One cooked food.” →
Treatment Procedures The instructional period lasted for seven consecutive, 50-min class meetings, with 25 min taken on Day 1 and Day 7 for testing. The control group likewise took a pretest and immediate posttest 1 week apart, whereas instruction offered a mix of both whole-class and small-group interaction targeting unrelated grammar items. Their classroom input thus included only incidental tokens of se, given that no purposeful attempt was made to either avoid or include it in the instructors’ speech. Although all learner groups had been introduced to se the previous semester during a lesson on reflexive pronouns (e.g., Example 1a), at the time of this study they had not yet learned of the anticausative se shown in Examples 1b and 1c, or its L1 equivalents, including the problematic zero derivation for the English middle voice (Example 1c).
The number of class meetings dedicated to
151 Language Learning 61:Suppl. 1, June 2011, pp. 141–188
Toth
Teacher- and Learner-Led Discourse
Table 1 Lesson topics for instruction on Spanish se Day Day 1 (Monday) Day 2 (Tuesday) Day 3 (Wednesday) Day 4 (Thursday) Day 5 (Friday) Day 6 (Monday) Day 7 (Tuesday) Lesson topics Administer pretest, review reflexive se with remaining time. Introduce and practice impersonal se. Introduce and practice passive se. Introduce and practice middle voice se of “unplanned occurrences.” Practice middle voice se with indirect object pronouns. Introduce and practice verbs of emotion with se. Review week’s lesson, administer immediate posttest.
se for this study was normal for the second-semester curriculum, with the only difference being that se topics were grouped together into a single week rather than being dispersed throughout the semester. Table 1 indicates the sequencing of se topics for the treatment groups. Here, it can be seen that after first reviewing reflexive and reciprocal uses, four class meetings dealt exclusively with the anticausative. Day 6 addressed se with psychological changes of state, which, due to the unique semantic features of these verbs, was presented in a separate publication (Toth, 2003).
Teachers for each group underwent separate 2-hr training sessions on how to implement instruction before beginning the treatment period. The lesson plans were designed and distributed to teachers the day before each class. For both treatments, each lesson began with a 5–10 min, whole-class “warm-up” during which teachers elicited multiple answers to one or two open-ended questions requiring the previous day’s use of se, as well as vocabulary items that had been assigned for that day’s lesson. The warm-up was then followed by 5–10 min of explicit, metalinguistic information on the current day’s use of se, which was identical for both treatments in order to confine their differences to the discourse format of subsequent activities. The grammar presentations first outlined the formation of sentences using se and then contrasted the formmeaning relationships signaled by its presence in detransitivized sentences versus its absence in transitive ones. English equivalents were given to identify the anticausative meaning in the examples, and learners were informed about typical errors that English speakers make.4 The remaining 30–35 min of each class involved either TLD or LLD activities requiring that day’s use of se, according to each treatment. In both groups, this time was generally organized around the completion of two communication tasks, such as describing and comparing pictures, putting a series
Language Learning 61:Suppl. 1, June 2011, pp. 141–188 152
Toth
Teacher- and Learner-Led Discourse
of events in order, or exchanging short personal narratives. The tasks were performed as information gap activities in groups of two or three for the LLD treatment and as whole-class collaborative activities for TLD. Some of these activities were designed specifically for this project; others came directly from the regular course textbook (Terrell, Andrade, Egasse, & Mu˜ oz, 1994).
Both n treatments were then assigned the same homework, which usually included reading a grammar explanation about the next day’s target item from the textbook, studying assigned vocabulary, and completing two to four accompanying written exercises, which occasionally involved short readings. In the LLD group, implementation of each day’s tasks involved brief pretask planning and posttask follow-up phases, both of which were conducted as whole-class interactions bracketing the small-group context in which the task itself was carried out. The pretask routine had teachers distributing necessary materials, informing learners of task objectives, and then asking them to brainstorm vocabulary items and interrogative expressions necessary for conducting the task in the L2. Answers to the brainstorm were often tabulated on the blackboard for reference during task performance. For the posttask activity, learners usually reported their results back to the whole class in order to receive feedback and confirm accuracy. Meanwhile, the task itself consistently required a two- or three-way exchange of information, so that each learner had to produce meaningful language and negotiate for mutual comprehension. During this time, LLD teachers circulated around the class, providing ad hoc assistance upon request. On average, LLD learners spent about 20 min total per lesson interacting in pairs or small groups to perform the two main tasks. Although one might argue that the time spent in groups could have been greater, learner peers were nonetheless the exclusive source of linguistic support during the tasks, which served as the principal points of contrast between the two groups. Figure 1 provides an illustration of the picture description task used on Day 4 to target anticausative se for describing physical changes of state. Here, learners were told that a hurricane had caused considerable damage to two similar-looking houses and that they had to identify the differences between them, with one person looking only at Picture A and the other looking only at Picture B. A pretask vocabulary brainstorm and posttask answer check bookended the task. The following day, learners worked with eight numbered pictures depicting a series of “unfortunate events” that happened to characters in their textbook, all of which required anticausative se plus an indirect object pronoun. One learner in each pair received the odd-numbered pictures in the sequence and the other received the even-numbered ones. Together, they then had to
153 Language Learning 61:Suppl. 1, June 2011, pp. 141–188
Toth
Teacher- and Learner-Led Discourse
Figure 1 A picture comparison information exchange activity targeting middle voice se.
reconstruct the sequence by describing each picture and pooling their information. Answers were confirmed in the posttask follow-up. The second task then engaged learners in a structured interview, during which they told each other of any accident like those in the first task that might have recently happened to them. After first rehearsing questions and follow-ups in a pretask activity,
Language Learning 61:Suppl. 1, June 2011, pp. 141–188 154
Toth
Teacher- and Learner-Led Discourse
partners had to ask for three additional details and take notes before reversing roles. The posttask activity then required them to write a short summary of their partner’s accident based on the exchange. In this way, learners moved from controlled picture descriptions to more open-ended personal narratives over the course of Day 4 and Day 5. Meanwhile, activities in the TLD group were similar to those of LLD, except that they were conducted as whole-class interactions rather than in small groups. On Day 4 and Day 5, for example, TLD learners used the same materials as LLD learners to describe the hurricane damage in Figure 1 and the “unfortunate events” of their textbook characters, except that all TLD learners could see both halves of the visual information and the tasks were implemented by having the teacher call on volunteers. Thus, at the beginning of each, teacher-initiated questions signaled the communicative goal to be accomplished (e.g., “D´ganme, ı e o ¿qu´ ocurri´ en el dibujo?” [Tell me, what happened in the picture?]), whereas subsequent initiation moves repeatedly referred back to the task purpose in order to elicit responses from other learners (e.g., “¿Qu´ m´ s pas´ ?” [What else e a o happened?]).
Often, like the pretask activities of the LLD group, a brainstormed list of necessary vocabulary was put on the board before engaging in the task itself. Because there was no need for posttask feedback, there was often time for one or two short, additional activities. For example, on Day 5, after describing their recent accidents to the class, TLD learners had time for another activity in which they used anticausative se to think up unfortunate events that would make good excuses for not coming to class. Thus, whereas individual LLD learners undoubtedly had more opportunities for L2 production in their groups, TLD learners performed a greater number of tasks due to lighter time constraints. To summarize, apart from time for additional activities in TLD, the principal points of contrast between the two treatments were found in the interactional features of the primary tasks. Both groups engaged in the same warm-up activities and received the same explicit grammar information; both engaged in pretask brainstorming; and both performed nearly identical meaning-oriented tasks requiring use of the target structure. Although LLD afforded greater opportunities for output to all learners, teachers rather than learner peers played a more prominent role in providing input and directing attention to form in TLD. Data Gathering and Assessment Assessment consisted of written production and grammaticality judgment tasks administered immediately prior to (pretest), immediately following (posttest),
155 Language Learning 61:Suppl. 1, June 2011, pp. 141–188
Toth
Teacher- and Learner-Led Discourse
and 24 days after instruction (delayed posttest) during testing periods that lasted 25 min total for each administration. In addition, Day 4’s lesson was videotaped in both treatments and selected portions transcribed. In each of the two LLD classes, one group of learner volunteers had their small-group interactions recorded on a portable audio recorder. Afterward, during the time between the posttest and delayed posttest, se appeared incidentally in the teachers’ input as it does in natural speech, but it was no longer the object of instruction. As mentioned previously, the control group took the tests at the same time intervals as the experimental groups, but with no intervening instruction on se. Native speakers took the same tests as learners, but only once. Before being administered to learners, test items were piloted on two Spanish native speakers to ensure their validity as assessments of targetlike performance. To control for test learning effects, two separate versions of the test were designed and distributed evenly and randomly among learners for the pretest. On the posttest, the version the learners received was switched to the one they had not yet taken. For the delayed posttest, the versions were switched again so that learners took the same one they had taken as a pretest over 1 month prior. As with the learners, when the native speakers took the test, both test versions were distributed randomly and evenly. Although each version had completely different items, the tasks were identical in overall length and in the number of items targeting anticausative se. An equal number of distracter items also appeared on both test versions. All words chosen for the test items came from a list of vocabulary that had appeared prominently in the Spanish curriculum prior to the beginning of the study. When taking the test, learners were instructed to ask for any vocabulary help they needed, although few actually did so. In order to prevent performance on the production task from being influenced by GJ task items, the production task was administered first. This consisted of 12 drawings accompanied by a verb and an NP, as shown in Figure 2, where the target response would have been “El vaso se rompi´ (The o glass broke).” Learners were asked to write a description for each picture using the words given. Four items were meant to elicit anticausative se, with two consisting of verbs representing inchoative events (Example 1c) and two consisting of externally caused ones (Example 1b).5 The other eight items on the task targeted se with verbs of emotion and cases where the use of se would have been ungrammatical, as reported in Toth (1997, 2000, 2003).
For all anticausative items, the given NP represented a patient undergoing the action of the verb. A reliability analysis of all learners’ posttest responses for these four items yielded Cronbach’s α values of .76 for Version A of the task, and .81 for
Language Learning 61:Suppl. 1, June 2011, pp. 141–188 156
Toth
Teacher- and Learner-Led Discourse
Figure 2 A sample item on the production task targeting middle voice se.
Version B. Meanwhile, a t-test checking for differences between the posttest means for the two versions found none, t(75) = 1.48; p = .14. Meanwhile, the GJ task asked learners to rate the acceptability of 50 sentences on a Likert scale from −3 for “very bad,” through 0 for “not sure,” to +3 for “very good,” as shown in Example 2.6 The instructions explicitly asked learners not to analyze the sentences, to move quickly through each item, and to answer intuitively according to how correct the sentence “felt.” The items to be considered here were eight sentences representing grammatical uses of anticausative se, including three middle-voice constructions and five impersonal or passive sentences.7 The Cronbach’s α value for all learners’ posttest responses on Version A was .83 and for Version B it was .73, with no differences between the versions found on a t-test, t(76) = −1.31; p = .19. The other 30 items on the task assessed overgeneralization, L1 transfer errors, and se’s use with verbs of emotion. Meanwhile, 12 distracters pertained to features of Spanish grammar unrelated to se. Again, a full report of these results can be found in Toth (1997, 2000, 2003).
Example 2. Sample items from the grammaticality judgment task: No se fuma aqu´. ı −3 −2 −1 0 +1 +2 +3 Se descompusieron las m´ quinas. −3 −2 −1 0 +1 +2 +3 a Se trae cerveza a todas las fiestas. −3 −2 −1 0 +1 +2 +3 (One doesn’t smoke here.) (The machines broke down.) (Beer is brought to every party.) Finally, upon recording the class interactions on Day 4, the spottingdifferences task shown in Figure 1 was chosen for transcription as a prototypical
157 Language Learning 61:Suppl. 1, June 2011, pp. 141–188
Toth
Teacher- and Learner-Led Discourse
example of an information-gap task that would best exemplify the contrasts between the two treatment groups. In accordance with Research Question 3, discourse segments were selected to shed light on the way anticausative se’s form-meaning relationships were attended to by the participants.
Scoring and Coding Procedures for the Production and GJ Tasks Because items on both tasks had been randomly ordered, those pertaining to anticausative se were first separated from the others to determine the participants’ scores. For the production task, given that the aim was to quantify use of the anticausative, answers that correctly used se in an intransitive sentence were awarded one point (e.g., for Figure 2, “Se rompi´ el vaso” [The glass o broke]), whereas all other answer types received a zero, even if they were grammatically acceptable alternatives. Thus, although some, including the native speakers, wrote grammatical transitive sentences without se at times (e.g., “Yo romp´ el vaso” [I broke the glass]), these answers received the same zero ı score as ungrammatical intransitives without se (e.g., “El vaso rompi´ ” [The o glass broke]) in order to facilitate assessment of the instructional object with inferential statistics. Likewise, errors pertaining to other features of Spanish morphosyntax, such as person-number or tense-aspect suffixes, were ignored unless they interfered with assessing the sentence’s transitivity, in which case they too received a zero. In this way, the production task scoring isolated grammatical uses of se in intransitive sentences from all other possible answers. After coding and scoring all responses, an average for each participant was determined by dividing the points earned by the four total anticausative items. Overall participant scores thus ranged from zero for no intransitive uses of se to a maximum of 1.00 for four uses, with intermediate scores representing the percent of intransitive se answers that the individual had provided. Having determined individual scores in this way, an analysis of variance (ANOVA) was performed as described in the Results section. On the GJ task, as on the production task, scoring involved first separating out the anticausative se items. The numbers used by learners to indicate their judgments were entered as raw scores into calculations of individual means for the eight grammatical items. Thus, scores ranged from −3 to +3. If a zero was given for any item, indicating an answer of “don’t know,” this data point was treated as a nonresponse and removed, so that individual means of zero would only come from averaging positive and negative answers. As on the production task, individual scores were then entered in the ANOVA presented in the Results section.
Language Learning 61:Suppl. 1, June 2011, pp. 141–188 158
Toth
Teacher- and Learner-Led Discourse
Table 2 Quantitative group results for anticausative se Task and statistic Production task Group mean Stand. deviation GT task Group mean Stand. deviation Control Learner-led Teacher-led
Pre Post Delay Pre Post Delay Pre Post Delay Native 0.01 0.00 0.05 0.00 0.71 0.61 1.18 1.31 0.03 0.11 0.80 0.85 0.02 0.48 0.08 0.32 0.91 1.74 0.94 0.64 0.33 0.30 1.27 0.99 0.04 0.63 0.10 0.32 0.86 2.17 1.01 0.66 0.50 0.34 1.93 0.71 0.48 0.20 2.24 0.75
Results Production Task The middle rows of Table 2 show group means and standard deviations for using anticausative se with intransitive syntax on the three test administrations. The means are also plotted in Figure 3, with native-speaker (NS) results included for comparison. From the graph it is clear that whereas NSs wrote intransitive se answers for roughly half the anticausative items (M = 0.48),8 the TLD group greatly exceeded this score on the immediate posttest (M = 0.63) and then reached a similar score on the delayed test (M = 0.50), albeit with considerably greater variation, as their larger standard deviations indicate. Meanwhile, the LLD group nearly reached the NS mean on the immediate posttest (M = 0.47) but then decreased somewhat on the delayed test (M = 0.33), with standard
Figure 3 Production task group means for use of anticausative se in intransitive sentences.
159 Language Learning 61:Suppl. 1, June 2011, pp. 141–188
Toth
Teacher- and Learner-Led Discourse
deviations on both posttests similar to those of TLD. By contrast, the control (C) group means and standard deviations remained close to zero throughout. A repeated-measures ANOVA with one between-subjects variable (treatment group) and one within-subjects variable (time of test) was performed on the three learner groups’ scores using a General Linear Model. This yielded statistically significant results for the interaction of treatment group with time of test, F(4, 148) = 15.20, p .05, η2 = 0.02, but on the two posttests p there were: for the immediate posttest, F(2, 74) = 40.73, p
GJ Task The lower two rows of Table 2 display group means and standard deviations for the acceptance of anticausative se on the GJ task over the three test
Language Learning 61:Suppl. 1, June 2011, pp. 141–188 160
Toth
Teacher- and Learner-Led Discourse
Figure 4 GJ task results for learner groups’ acceptance of anticausative se.
administrations. Group means are also plotted in Figure 4, with the NS result given for comparison. Here, it is clear that both instructed groups improved markedly from the pretest to posttest, with some decrease on the delayed posttest, whereas the C group remained nearly unchanged. From pretest to immediate posttest, the TLD mean increased by 1.31, nearly reaching that of the NS group, whereas the LLD mean increased by 0.83 to within a half-point of the NS mean (posttest: TLD = 2.17, LLD = 1.74, NS = 2.24).
Changes from the immediate to delayed posttest were greater for LLD than for TLD, with LLD decreasing by 0.47 and TLD by 0.24. From pretest to delayed posttest then, TLD gains were greatest at 1.03, followed by LLD gains at 0.36, and C group gains 0.09. Of further interest is the change in standard deviations among the three groups over time. For the C group, standard deviations on all three tests were considerably higher than the NS group’s 0.75, indicating more variation among C group individuals than among NSs. Standard deviations for the two instructed groups were also greater than NSs on the pretest (LLD = 0.94, TLD = 1.01) but became smaller on the immediate posttest (LLD = 0.64, TLD = 0.66), indicating that individual scores were closer to the group means. By the delayed posttest, however, the LLD standard deviation returned to the higher pretest value (0.99), whereas the TLD group remained lower than the NS result and closer to the immediate posttest outcome (0.71).
Thus, in addition to a lower group mean for LLD on the delayed posttest, individual scores fell farther from that mean in LLD than in TLD. This greater variation in LLD scores suggests that whereas the instructional benefits evident on the immediate posttest might
161 Language Learning 61:Suppl. 1, June 2011, pp. 141–188
Toth
Teacher- and Learner-Led Discourse
have persisted for some in that group, they did not do so for as many as in the TLD group. As on the production task, a General Linear Model repeated-measures ANOVA was performed on the GJ data, with treatment group as the betweensubjects variable and time of test as the within-subjects variable. Statistically significant results were found for the interaction of treatment group with time of test, F(4, 150) = 6.45, p .05, η2 = 0.01, p but significant differences were found on both the immediate posttest, F(2, 75) = 20.08, p .10).
Thus, on the GJ task, as on guided production, a case can be made for beneficial effects attributable to both treatments. However, after instruction, the TLD group showed a stronger performance on both tasks, and where grammaticality judgments are concerned, TLD learners appeared to have retained the benefits of instruction more consistently than did their LLD counterparts. Discourse Transcripts Excerpts from the two instructed groups’ performance of Day 4’s picture description task (Figure 1) are presented here, with the transcribed segment starting roughly at the beginning of the interaction. As per Research Question 3, the discussion of each excerpt focuses on the learners’ use of anticausative
Language Learning 61:Suppl. 1, June 2011, pp. 141–188 162
Toth
Teacher- and Learner-Led Discourse
se as well as their feedback and comments on L2 form-meaning relationships. Throughout the transcriptions, English equivalents are given in parentheses and nonverbal gestures, as well as errors and relevant L2 linguistic information, are given in brackets. A zero with a slash (Ø) indicates an incorrect omission of se; three periods indicate extended pauses; an extended dash indicates an interrupted utterance; and all names appearing in the text have been changed to aliases. The LLD discourse samples below come from one of the two course sections, where, after dividing an odd number of individuals into pairs, it was the one group comprised of three learners that volunteered to have their interaction recorded.10 As a result, in Example 3, the learners began with Student 1 having picture A in Figure 1, and Students 2 and 3 sharing picture B. From their interaction, it is clear that despite instructions to the contrary, all three were holding the pictures so that their partners could see them, and Student 2 seemed to believe that the sofa stain in Student 1’s picture came from the wine bottle on the floor. Example 3 a. Student 1: S´, preg´ ntame, Ra´ l, de mi foto. ı u u (Yes, ask me, Ra´ l, about my picture.) u b. Student 2: Ah, en tu dibujo, ¿Qu´ le pas´ el botalla? [pronunciation and e o gender error] (Ah, in your picture, What happened the bottle?) c. Student 1: ¿Botalla? [again, pronounced incorrectly] (Bottle?) d. Student 2: ¿de vino? (of wine?) e. Student 1: Oh, nadie le pas´ . [lexical error using nadie to mean o “nothing”] (Oh, no one [sic] happened to it.) f. Student 2: vino en— [indicating the wet spot on the sofa] (wine on—) g. Student 1: Oh, ok. El vino, la botella se moja la sof´ . [errors w. se and a gender of sof´ ] a (Oh, ok. The wine, the bottle [se] wets the sofa.) h. Student 2: ¿Moja o mojo? (Wets or wet?) i. Student 1: Se moj´ . Like it wet the couch. So, la botella se moj´ la sof´ . o o a ([Se] wet. Like it wet the couch. So, the bottle [se] wet the couch.)
163 Language Learning 61:Suppl. 1, June 2011, pp. 141–188
Toth
Teacher- and Learner-Led Discourse
j. Student 3: O se cay´ . o (Or it fell down.) k. Student 1: ¿Se cay´ ? o l. Student 3: Se cay´ . o m. Student 1: But it didn’t break the couch. n. Student 3: No, caer. It fell. Here, it is clear that during the task, learners were indeed focusing on, negotiating over, and proposing hypotheses for a number of necessary L2 forms. Specifically, in line h, Student 2 questioned Student 1 about the correct past tense suffix for the verb mojar (to wet), and in lines j–n, Student 3’s proposal to use the verb caer (to fall) resulted in a negotiation with Student 1 over the verb’s meaning. Clearly, this provides evidence that LLD yielded facilitative L2 “metatalk” (Swain, 1998, 2000) while also supporting Pica et al.’s (1993) claim that information gap tasks are well suited to trigger negotiation. However, despite the learners’ considerable attention to L2 morphosyntax and lexis, it is noteworthy that in lines g and i, Student 1’s incorrect use of se with the transitive form of mojar failed to capture attention, as her peers’ concerns centered on correct past tense forms. Although not shown here, preoccupation with the past tense was also evident later in the recording, where Student 2 initiated another round of negotiation over the possibility that the vowel /e/ in the verb cerrar (to close) might change to a diphthong when inflected in the past, as indeed it does in the present. Although Student 1 then played an important role in resolving this uncertainty, se often appeared to be an “unanalyzed chunk” in her spontaneous speech throughout the recording, not unlike the clitic pronouns documented by Myles, Mitchell, and Hooper (1999) among beginning L2 French learners. Nonetheless, a few moments later, Student 1 demonstrated that with peer assistance, she could apply instruction-derived metalinguistic knowledge to reanalyze one of her se + verb chunks and correct the overgeneralization error, which this time involved the verb entrar (to enter).
Example 4 a. Student 2: ¿C´ mo se dice “wind”? o (How do you say “wind”?) b. Student 1: Viento. Ok, you could say, maybe, “Se, se entra [overgeneralized se] viento en la casa,” like— (Wind. Ok, you could say, maybe, “Wind [se], [se] enters the house,” like—) c. Student 3: Se, uh, entr´ , ¿verdad? o
Language Learning 61:Suppl. 1, June 2011, pp. 141–188 164
Toth
Teacher- and Learner-Led Discourse
(It [se], uh, entered, right?) d. Student 1: Se entr´ viento en la casa. o (Wind [se] entered the house.) e. Student 3: Se entr´ . o (It [se] entered.) f. Student 2: Oh, se— ¿Se entr´ ? o (Oh, [se]— It [se] entered?) g. Student 1: Se entr´ . Oh, but would you use se with that? o h. Student 2: You wouldn’t. i. Student 1: No, because that’s a subject. Right. Viento entr´ , entr´ en la e o casa (Wind enter, entered the house.) j. Student 3: S´. ı (Right.) Here again, it is evident that a major preoccupation for the learners was the past tense, as in line c, where Student 3 questioned Student 1’s use of the present in line b. An overgeneralized se also appeared in this utterance and was repeated by Students 1 and 3 in lines d and e before Student 2 detected a possible problem in line f. This then led Student 1 to reconsider her use of se in line g and, with support from Student 2 in line h, apply her own explicit, metalinguistic understanding to correct the overgeneralization in line i. This dynamic is thus reminiscent of the small-group interactions in Donato (1994), in which collaboration among peers yielded linguistic outcomes that exceeded the limitations that any one of them would have faced alone. Given this evidence of facilitative interactions for L2 grammatical development, the question then arises as to how TLD interactions might have differed to yield the stronger quantitative results observed here. Example 5 demonstrates how the picture description task in Figure 1 began in one of the two TLD classes. As previously mentioned, all learners were looking at pictures A and B, and the teacher started by asking them to describe differences between the pictures and then continued by soliciting multiple responses without changing the task focus. Example 5 a. Teacher: Venga, vamos a ver. ¿Qu´ ocurri´ durante la tormenta? ¿Qu´ e o e ocurri´ ? S´. o ı (Okay, let’s see. What happened during the storm? What happened? Yes?)
165 Language Learning 61:Suppl. 1, June 2011, pp. 141–188
Toth
Teacher- and Learner-Led Discourse
b. Student 4: Um, la ventana se, um abri´ . o (Um, the window, [se] um opened) c. Teacher: Bien. En el dibujo B, ¿verdad? En el dibujo B la ventana se abri´ . En el dibujo A, ¿qu´ ? o e (Good. In drawing B, right? In drawing B the window [se] opened. In drawing A, what?) d. Student 4: En el dibujo A, la ventana. . . (In drawing A the window. . .) e. Teacher: S´. . .¿Se abri´ ? No. Se. . . ı o (Yes. . .Did it [se] open? No. It [se]. . .) f. Student 4: No, se. . .ce-, ce-, cerr´ . o (No, it [se]. . .cl-, cl- closed.) g. Teacher: Perfecto. En el dibujo A se abri´ la ventana, y en el dibujo B— o No. En el dibujo A se cerr´ la ventana y en el dibujo B, se abri´ . o o ¿M´ s? Jim. a (Perfect. In drawing A the window [se] opened and in drawing B—No. In drawing A the window [se] closed and in drawing B it [se] opened. What else? Jim.) h. Student 5: Um, en B, la luz, uh, ca-, uh, cay´ . [incorrect omission of se] o (Um, in B, the light, uh, f-, uh, [Ø] fell.) i. Teacher: S— [making a falling gesture with her hands, and then pointing toward another learner with his hand up.] j. Student 6: Se cay´ . o (It [se] fell down.) k. Teacher: Perfecto. (Perfect.) l. Student 5: Se cay´ . o (It [se] fell down.) m. Teacher: Se cay´ . La luz se cay´ . o o (It [se] fell down. The light [se] fell down.) Thus, following the teacher’s elicitation in line a, Student 4 successfully used se in line b with the verb abrir (to open).
The teacher’s feedback in line c affirmed the answer and repeated it to the class before prompting Student 4 to expand her utterance and comment on the second picture. When Student 4 hesitated in line d, the teacher’s response in line e was to further model se in a false statement (se abri´ ) and then assist her by starting the target sentence with se. o After the learner was able to complete the utterance in line f, the teacher then summarized what had been said so far in line g and elicited another response.
Language Learning 61:Suppl. 1, June 2011, pp. 141–188 166
Toth
Teacher- and Learner-Led Discourse
When Student 5 incorrectly omitted se from his answer in line h, the teacher’s feedback in line i consisted simply of the sound of an “s” accompanied by a hand movement. Student 6 then volunteered a correct answer in line j, which was repeated by Student 5 in line l and the teacher in line m. Unlike the LLD group, this example reveals that the teacher’s participation produced a recognizable IRF pattern, with utterances in lines c and g consisting of feedback for the preceding learner turn, followed by subsequent initiation moves. However, a number of the lines in between, such as lines e and i, seemed to serve the additional function of assisting with utterance formulation rather than merely evaluating performance (i.e., feedback) or posing an additional question and nominating a speaker (i.e., initiation).
This kind of assistance is somewhat different from typical characterizations of IRFbased instructor “scaffolding” because it is not designed to shape the topical content of discourse or engage analytic abilities, as initiation and feedback moves often do (e.g., Cazden, 2001; Tharp & Gallimore, 1988; Wells, 1993; Wood, Bruner, & Ross, 1976).
Instead, lines e and i appear uniquely intended to lighten the cognitive load that mapping the target L2 form onto meaning entails during production, particularly when learners have insufficient control over necessary forms. Although there was also ample evidence of assistance in working out L2 form-meaning relationships among peers during LLD, it did not occur as direct procedural support for articulating L2 utterances, as in TLD. As the task continued, Example 6 suggests that the teacher’s ongoing assistance in formulating targetlike utterances with se led some learners to monitor their speech and attempt to preempt errors on their own. Example 6 a. Teacher: Muy bien. ¿Qu´ m´ s? Luis. e a (Very good. What else? Luis.) b. Student 7: Dibajo A [both mispronounced] la puerta— (Drawing A the door—) c. Teacher: ‘A’ [correcting his pronunciation of the letter A] d. Student 7: ‘A’, la puerta, uh, a-, ¿abri´ ?. . .Se, se abri´ . o o (‘A’ the door, uh, [Ø] o-, opened?. . .It [se] opened.) e. Teacher: S´, la puerta se abri´ . ı o (Yes, the door [se] opened.) f. Student 7: En dibujo B, la puerta cerr´ . [incorrect omission of se] o (In drawing B, the door [Ø] closed.) g. Teacher: S—
167 Language Learning 61:Suppl. 1, June 2011, pp. 141–188
Toth
Teacher- and Learner-Led Discourse
h. Student 7: [simultaneously] Se cerr´ . o (It [se] opened.) i. Teacher: Perfecto. La puerta se cerr´ . o (Perfect. The door [se] closed.) Thus, in line d, as Student 7 began monitoring his output, he was able to detect his L1-derived tendency to omit se with the verb and correct the error. However, when it appeared that he would omit se in line f, the teacher’s immediate assistance in line g—again, with the sound of an “s”—led to simultaneous self-correction and a targetlike utterance in line h. Again, it appears that the teacher’s support helped him not only to formulate his utterance but also to develop the procedural self-regulation necessary for overriding an L1 transfer error. In addition to correcting omission of the object of instruction, Example 7 demonstrates how teacher feedback and assistance also led learners away from overgeneralizing se. It is indeed noteworthy that Student 8’s error with the verb mojar in this example is nearly identical to that of Student 1 in the LLD group, discussed in Example 3. Example 7 ı a. Teacher: S´. Susana. (Yes, Susana.) b. Student 8: Um, en dibujo A, um, la lluvia. . . (Um in drawing A, um, the rain. . .) c. Teacher: . . .Mm-hm. . . d. Student 8: . . .¿Se? [overgeneralization error] (. . .got?) e. Teacher: No s´ . Venga, venga. La lluvia ¿qu´ ? e e (I don’t know. Come on, come on. The rain what?) f. Student 8: moj´ . o (wet) g. Teacher: ¿La lluvia se moj´ ? o (The rain got wet?) h. Student 8: Mm-hm. i. Teacher: ¿La lluvia se moj´ [making a hand motion toward herself]? Eso o es muy metaf´sico. Eso es muy, como, misticismo. ¿C´ mo que ı o la lluvia—? La lluvia no se moj´ [again, motioning toward o herself]. La lluvia, a s´ misma. ı
Language Learning 61:Suppl. 1, June 2011, pp. 141–188 168
Toth
Teacher- and Learner-Led Discourse
j. k.
l. m. n. o.
p. q.
(The rain got wet? That is very metaphysical. That’s very, like, mysticism. How could the rain—? The rain didn’t get itself wet. The rain, to itself.) Student 8: ¿Moj´ ? Moj´ . o o It wet? It wet. Teacher: La lluvia moj´ [motioning as if putting the sentence together o e linearly] ¿qu´ ? (The rain got what wet?) Student 8: ¿C´ mo se dice carpet ? o (How do you say “carpet”?) Teacher: ¿C´ mo se dice? o (How do you say it?) Student 9: El alfombro [incorrectly marked as masculine noun class] (The carpet) Teacher: Alfombra. Muy bien. La alfombra. La lluvia moj´ la alfombra. o O ¿la alfombra. . .? (Carpet. Very good. The carpet. The rain got the carpet wet. Or the carpet. . .?) Student 8: Se moj´ . o (Got wet.) Teacher: Perfecto. La alfombra se moj´ . o (Perfect. The carpet got wet.)
Thus, as Student 7 began to formulate her utterance in lines b and d, the teacher provided a supportive syntactic frame in line e to indicate a missing element (“La lluvia ¿qu´ ?”).
When the learner completed her sentence with “se moj´ ” e o in line f and confirmed in line h that she indeed meant to say it, the teacher then gave metalinguistic feedback in line i to demonstrate the anomalous reflexive that had been created. When Student 7 then proposed the same verb without se in line j, the teacher again offered procedural support, this time for a transitive sentence, by repeating the utterance to that point and adding an interrogative “¿qu´ ?” to indicate the slot that had yet to be filled. When a vocabulary problem e emerged in line l, the teacher then posed the question to the whole class, with another learner responding in line n but making an error in the noun’s gender. Without detracting attention from the instructional object, the teacher in line o simply recast the noun with correct gender to complete the transitive sentence. Then, to confirm knowledge of when se is and is not required, the teacher posed the beginning of an intransitive sentence to Student 7, who in line p then used se in a targetlike manner.
169 Language Learning 61:Suppl. 1, June 2011, pp. 141–188
Toth
Teacher- and Learner-Led Discourse
In Examples 5, 6, and 7, the teacher’s discourse leadership appeared to have yielded a more consistent focus on the object of instruction than the LLD learners’ discussion of L2 forms, which embraced a broader gamut of structures. The TLD instructor’s initiation moves provided targetlike models of se and elicited focused learner output, whereas assisting moves lightened the procedural burden of mapping L2 form onto meaning and reminded learners of targetlike sentence patterns. Finally, feedback moves either confirmed the grammatical hypotheses underlying learners’ responses or gave disconfirming evidence requiring a reformulation of their utterances. Often, they also provided elaborated input demonstrating a key grammatical contrast. In LLD however, with no one designated as an L2 expert, attention was directed to L2 forms via questions, feedback, and metatalk, without the singular focus on one linguistic issue and with greater negotiation among participants. In addition, without a manager of turn-taking and elicitation, the available participant roles did not seem to allow direct assistance to others in articulating their utterances, as the instructor role did in TLD. This final excerpt from the three LLD learners aptly demonstrates this point, as, toward the end of the interaction, Student 3 sought assistance from Student 1 in saying that the couch in his drawing had moved. Although Student 1 had begun to emerge as a relative expert among the participants, the assistance provided nonetheless required negotiation and did not directly push Student 3 to formulate his intended utterance, as Student 8 did in Example 7 during TLD. Example 8 a. Student 3: ¿Qu´ es la palabra para desde aqu´ a all´? e ı ı (What is the word for from here to there?) b. Student 1: “Derecha.” (“Right” [i.e., the opposite of “left”]) c. Student 3: “Derecha.” (“Right.”) d. Student 1: Mm-hm, verdad. (Mm-hm, correct.) e. Student 3: S´. ı (Yes.) f. Student 2: Hm. S´. ı (Hm. Yes.) g. Student 1: Derecha a izquierda. (Right to left)
Language Learning 61:Suppl. 1, June 2011, pp. 141–188 170
Toth
Teacher- and Learner-Led Discourse
h. Student 3: [simultaneously] ¿“Derecha” es el verbo? (“Right” is the verb?) i. Student 1: No, no, no. Derecha es el. . . (No, no, no. Right is the. . .) j. Student 3: The direction. k. Student 1: Yeah, right. Like derecha a izquierda. (Yeah right. Like right to left.) l. Student 3: ¿Qu´ es, ah, el, qu´ es el verbo? e e (What is, ah, the, what is the verb?) m. Student 1: [to Student 2] ¿Tus padres hablan espa˜ ol? n (Do your parents speak Spanish?) n. Student 2: S´. ı (Yeah.) o. Student 3: [to Student 1] ¿Qu´ es el verbo? e (What is the verb?) p. Student 1: Oh, para—mojar, mojar, mojar. Er, no, no, no. Moverse. That’s it. (Oh, for—to wet, to wet, to wet. Er, no, no, no. To move. That’s it.) [Student 3 starts spelling moverse aloud, and the other assist him, letter by letter.] q. Student 2: Se movi´ la sof´ — o a (The sofa [se] moved—) r. Student 1: Se movi´ — o (It [se] moved—) s. Student 2: a la derecha. (to the right.) t. Student 1: a la derecha. Muy bien, Ra´ l. Eres muy inteligente. . .Hoy! u [laughs]. (to the right. Very good, Ra´ l. You’re very intelligent. . . u Today!) Although this excerpt took up no more than 40 s in real time, the 10 turns in lines b–k clearly occurred due to a misunderstanding of what Student 3 intended to ask. Although such misunderstandings could obviously occur in TLD as well, the difference between Student 1’s role as relative authority and that of the instructor in TLD is evident in that Student 1 at no point took responsibility for making Student 3 produce his sentence. In other words, Student 1 never sought to confirm Student 3’s learning by prompting him to use the new word
171 Language Learning 61:Suppl. 1, June 2011, pp. 141–188
Toth
Teacher- and Learner-Led Discourse
in a meaningful sentence, as the teacher in Example 7 did when checking that Student 8 could use se intransitively. Indeed, Student 3’s discovery in line h that he had the wrong word apparently occurred by accident after Student 1 in line g provided an unprompted amplification of her previous utterance. By the time Student 3 gained clarification of the meaning of “derecha” in line i and line k, his request for the verb he wanted had to be made twice, in line l and line o, given that Student 1’s attention had momentarily shifted from the task to asking Student 2 about his family. Although in line q and line s, Student 2 ultimately assembled the parts of the sentence that had thus far emerged, and in line t, Student 1 provided teacherlike feedback to Student 3, it is clear that the linguistic work of articulating Student 3’s intended sentence took place only after an extended negotiation not only for meaning but also for attention and without the online procedural assistance evident in TLD. Summary To review, on the quantitative assessments, the LLD and TLD groups both outperformed the C group for knowledge of Spanish se after a week of specialized instruction. However, regarding Research Questions 1 and 2, which asked whether LLD learners would outperform TLD learners, it appears the answer is “no” in both cases, given the TLD group’s higher postinstruction means on the production task and greater retention of gains on the GJ task. For Research Question 3, which inquired about qualitative differences in the way LLD and TLD learners attended to L2 form-meaning relationships, it appears that learners’ attention to form in LLD might have been more broadly focused on a number of linguistic issues than in TLD and that LLD participatory roles might not have allowed for the same procedural assistance in using the target form that was afforded to TLD learners. Discussion Taken together, the quantitative and qualitative data suggest that the teachercoordinated interactions in the TLD group might have better equipped learners for performance with the instructional object than the small-group learner-led interactions in LLD. Although the larger study (Toth, 1997, 2000, 2003) showed that L1-derived errors decreased similarly in both groups after instruction, and indeed the assessments here were not sophisticated enough to demonstrate a broader grammatical restructuring along the lines of the parameter shifts described by generative linguists (e.g., Bruhn de Garavito, 1999; White, 2003), what is clear is that the TLD learners’ more steadfast GJ gains and more
Language Learning 61:Suppl. 1, June 2011, pp. 141–188 172
Toth
Teacher- and Learner-Led Discourse
consistent, targetlike uses of se for picture descriptions coincided with qualitative evidence indicating more consistent appeals for attention to the morphosyntactic properties of se. Whether these results reflect a profound change in implicit linguistic knowledge or merely an improved explicit memory for instructed forms, the implications for L2 pedagogy are relevant, given an acquisition theory that views attention to L2 form-meaning relationships and any resulting explicit knowledge as at least facilitative of L2 grammatical development (e.g., DeKeyser, 2003; Doughty, 2001; Schmidt, 2001; Terrell, 1991).
Thus, despite the presumed more frequent learner turns in LLD and the optimal opportunities for modified output and negotiation that an information gap imparts, there was no advantage for this group when compared to TLD learners engaging in instructor-managed collaborative discourse. Given these results, what indeed can be said about TLD versus LLD in an L2 pedagogy that promotes morphosyntactic development?
Attentional Focus in LLD To the extent that teachers do not directly manage interactions in small learner groups, it seems fair to say that LLD is indeed more “learner centered” than TLD, which brings both strengths and limitations for target-structure learning outcomes that were apparent in this study. First, it is clear that, with greater possibilities for autonomy and self-selection in turn-taking, learners are indeed freer to focus attention on a broad array of topics, adopt a wide variety of speaking roles, and potentially utilize a greater number of linguistic structures during interaction. Clearly, despite the information-gap design, control over what to say and how to say it is looser. Although not systematically quantified here, a cursory glance at the three LLD transcripts does suggest many more learner-initiated questions and topics than in TLD and a broader array of L2 linguistic issues addressed. Nonetheless, despite ample evidence of collaboration and feedback among learners, the process of producing L2 utterances and reaching a consensus on their acceptability appears to have entailed more protracted negotiations in LLD than in TLD, with each participant having the right to inquire, hypothesize, and give opinions about a variety of issues. Although other research has shown important pedagogical benefits to be derived from such interactions (e.g., Donato, 1994; Foster & Ohta, 2005; Swain & Lapkin, 1995, 1998, 2001; van Lier, 1996), the data here suggest that this might have come at the expense of consistency in attention to the target form for the LLD group, given that requirements to use se were manifest only in the task instructions and design.
173 Language Learning 61:Suppl. 1, June 2011, pp. 141–188
Toth
Teacher- and Learner-Led Discourse
Although advocates for LLD as essential to task-based pedagogy would argue that such conditions ought to yield maximally “pushed” output and attention to targeted structures (e.g., Fotos, 2002; Loschky & Bley-Vroman, 1993; Long, 1996; Long & Porter, 1985; Pica, 1994; Pica et al., 2006), it appears that LLD participants here inevitably focused on nontargeted task-essential structures as well, most notably lexical items that might not have been sufficiently primed in pretask activities, and other morphosyntactic features that stood at the frontier of their current L2 linguistic competence (i.e., past tense morphology).
Thus, when assessments of se were administered to learners who had spent significant time directing attention more broadly to the L2 during self-initiated communication, their performance was not as robust as those who had spent equivalent time in teacher-assisted production using the target structure. It would therefore be reasonable to speculate that, given the considerable attention to L2 issues other than the object of instruction that the learners themselves selected as problematic, the assessments in this study might have failed to fully capture the instructional benefits derived from the LLD format. Indeed, a more openended assessment, or one that tracked learners more closely after instruction on the L2 structures that emerged in their metatalk, might have demonstrated stronger results. Although documenting learners’ posttask performance with issues emerging from LLD can indeed further our understanding of the relationship between learner-learner interactions and L2 development, confirmation of these benefits vis-` -vis TLD would require a comparison with TLD learners on a those same LLD-initiated issues as well as a teacher-selected object of instruction. In this way, one could discern whether the LLD-generated attention to L2 structures that are of concern to learners benefits morphosyntactic development in the same way as TLD-generated attention to structures that are of concern to the instructor. Optimal Teacher-Led Discourse, Attention to Form, and Procedural Assistance Turning to the TLD results, the transcript data suggest that this group might have experienced an optimal instantiation of this discourse format that avoided many of the potential pitfalls discussed in the literature. In particular, there was no evidence of the “local lexical chaining” in grammar practice described by Hall (1995, 2004) or any of the incoherently wandering closed questions documented by Lee (2000) and Leemann-Guthrie (1984) during “class discussions.” Instead, as per instructions, the teacher in Examples 5, 6, and 7 used links between the target structure and the task goals as the organizing feature for discourse (i.e., identifying differences between the two pictures) and avoided
Language Learning 61:Suppl. 1, June 2011, pp. 141–188 174
Toth
Teacher- and Learner-Led Discourse
the incongruent exchanges that would have resulted had she sought production of all the reflexive pronouns or all of the uses of se via “meaningful” and “communicative drills” (Paulston, 1972) that fail to cohere to a broader discourse purpose. In addition, elicitation via task instructions and open-ended rather than closed questions appeared to maximize learner output, given that natural responses entailed sentence-level rather than single-word utterances (e.g., “Describe the differences in these pictures” and “What happened in the picture?” rather than “Did the window open?”).
Finally, the solicitation of multiple learner responses before new questions were posed created a discourse space where several learners could experience task performance and feedback before subsequent questions should change the discourse focus. This feedback was optimal in that it provided elaborated subsequent input to learners and focused on key grammatical contrasts. Although the design of this study does not permit definitive causal links to be made between these qualitative observations and TLD’s quantitative results, their coincidence does suggest a facilitative impact on L2 development, given current views on the role of attention, scaffolded output, and language processing in acquisition. For example, one likely benefit of eliciting multiple learner responses for each teacher question is that previous teacher input and feedback would have been directly relevant to anyone speaking after the first respondent. In light of Ohta’s (2000, 2001) finding that TLD listeners often actively reflect on feedback to others, the benefit of being able take one’s own turn after that of others cannot be underestimated, as the value of instructor support becomes cumulative over the course of the task. Indeed, in Toth’s (2004) study, lower achieving L2 Spanish learners reported that they often depended on the previous answers of peers in order to both comprehend teacher questions and formulate appropriate responses. Such benefits are lost, however, when only one response is elicited before the topic/task changes to a new question, given that learners must then disregard as irrelevant the feedback provided for the first question and think up new appropriate answers for the second. Thus, the strategy of maintaining a single task/question’s relevance over several learner turns might have created an optimally cohesive context for engaging learners in meaningful production while assisting them in target structure use. Also, to the extent that non-turntakers were still cognitively engaged, the hypothesis-testing, gap-noticing, and metalinguistic functions of output proposed by Swain (1995, 1998, 2000) might have been maximally realized here as a collective mental effort with cumulative effects that would offset the fewer output turns per learner in TLD than in LLD.
175
Language Learning 61:Suppl. 1, June 2011, pp. 141–188
Toth
Teacher- and Learner-Led Discourse
Within this context, the qualitative data suggest that the teacher’s consistency in highlighting the instructional object was achieved in four important ways: (a) modeling se in her input, (b) cueing and eliciting it as output, (c) focusing feedback on ungrammatical omissions or overgeneralizations, and (d) actively assisting learners in articulating utterances. As noted in the transcripts, this latter goal was often accomplished via a sentence-building “fill-in-theblank” technique that provided a syntactic frame for linking words together while isolating missing or problematic elements in what had been said so far. These turns, which will here be dubbed “procedural assistance” moves, were often inserted between the lines of a typical IRF sequence as a unique feature of L2-oriented TLD, given that they were designed not to facilitate analysis of a concept or clarification of a previous utterance but to help the learner map L2 form onto meaning online, during utterance formulation. Although it is relevant to cite Wood et al.’s (1976) original exposition of the ways that teachers “scaffold” learners toward instructional goals via moves that simplify tasks and identify key performance features, the procedural assistance observed here is nonetheless unique to the L2 context in that it supports an articulatory task made complex by the linguistic structures involved rather than a conceptual one made complex by demands for critical thinking. Furthermore, this procedural assistance differs from the clarification requests, comprehension checks, recasts, and other feedback identified as useful for L2 development in small-group contexts, given that it is made directly to the learner while he or she formulates an utterance, rather than as an indication afterward that a communication breakdown has occurred. Instead, this assistance is more consistent with Johnson’s (1996) proposal that teachers might facilitate the development of fluent communication in learners through strategically targeted “hints” that support the “proceduralization” of subcomponents of the linguistic task (p. 108–109).
However, as noted previously, it is doubtful that interaction roles outside of TLD, whether among LLD peers or in nonclassroom settings, would allow interlocutors to offer such assistance. Indeed, Buckwalter’s (2001) finding that LLD participants strongly prefer self-correction to other-correction not only supports this conclusion but also concurs with findings from analysts of discourse in noninstructional settings (Seedhouse, 2004).
Thus, TLD might be a unique context for providing procedural assistance in L2 utterance formulation. Procedural Assistance and Output Processing If indeed L2 procedural assistance is licensed within TLD but generally dispreferred elsewhere, then how can they be theoretically justified as beneficial to
Language Learning 61:Suppl. 1, June 2011, pp. 141–188 176
Toth
Teacher- and Learner-Led Discourse
L2 acquisition? Recent work on instructed L2 development has proposed that pedagogical interventions might be most effective when they assist learners in language processing or, rather, pushing them to correctly encode specific formmeaning relationships in the L2 while using the language meaningfully (e.g., Carroll, 2001; Sharwood-Smith, 2004; VanPatten, 1996, 2004b).
Current models for how learners process L2 form-meaning links suggest that hierarchical semantic and syntactic relationships might be mentally computed online while comprehending or producing an utterance rather than after an utterance is concluded (Harrington, 2001; Pritchett, 1992).
Thus, the cognitive complexity of an utterance’s linguistic structure has been shown to affect communicative task performance (Juffs, 1998, 2004; Juffs & Harrington, 1995; VanPatten, 1990, 2004b).
To date, L2 processing research has focused primarily on how parsing crashes and/or misinterpretations of L2 input are triggered by the appearance of specific morphemes in the input, such as case-marked pronouns (VanPatten, 1990, 2004b) or inflected verbs in subordinate clauses (Juffs, 1998, 2004; Juffs & Harrington, 1995).
VanPatten has proposed “processing instruction” as a means of addressing these difficulties via explicit attention to troublesome L2 form-meaning relationships that appear in the input. Meanwhile, where output is concerned, current models of speech production have ascribed an important role to preverbal message “monitoring” as an accompaniment to lexical and grammatical encoding, such that the potential exists for explicit knowledge of L2 form-meaning relationships to affect the content of speech output (Izumi, 2003; Levelt, 1989; Scovel, 1998).
It follows, then, that when the structural complexity of one’s usable metalinguistic knowledge exceeds that of the extant implicit L2 grammar, L2 development might be helped if such knowledge can be employed online to guide lexical selection and grammatical encoding during communicative task performance (deBot, 1996; Kormos, 1999; Toth, 2006).
If through procedural assistance instructors facilitate not only the use of metalinguistic information but also the assembly of structurally more complex utterances to represent learners’ intended meanings, then such assistance might indeed be useful as a form of “pushed output” (Swain, 1985), given that part of the cognitive burden associated with the utterance’s linguistic structure is borne by the teacher and the production work left to the learner focuses only on those elements that pertain to the instructional object. The strongest support here for this idea appears in Example 7, where metalinguistic feedback first led Student 8 to retreat from overgeneralization and then the teacher’s procedural assistance helped her construct not only the intended transitive sentence but also an intransitive one correctly using the target form. In this way, not only
177 Language Learning 61:Suppl. 1, June 2011, pp. 141–188
Toth
Teacher- and Learner-Led Discourse
were contrastive examples of se’s use and nonuse provided—both as output for Student 8 and input provided to other learners—but the full burden of managing the syntax of each utterance as they emerged online was shared by both participants. Although further investigation is surely needed to confirm these ideas, it might be that such assistance in “output processing” constitutes an important initial step in developing the complexity of the learner’s linguistic repertoire. In making a case for output during LLD-based collaborative tasks, Swain (2000) has asserted the possibility that “output pushes learners to process language more deeply—with more mental effort—than does input” (p. 99), especially given the learner’s ability to select and encode his or her own intended meaning. The benefits of such autonomy notwithstanding, it might be that although there are still fewer turns per learner in successful TLD, a collective mental effort in processing output is nonetheless achieved, as teacher prompts and feedback focus learners’ attention on the L2 target and a sense that all are sufficiently enfranchised to take a turn makes nonparticipants into engaged listeners when others are taking theirs. Limitations of the Study and Directions for Future Research Although this study has shown that L2 grammar instruction given as TLD can yield stronger learning outcomes than LLD for target forms, it clearly should not to be taken as evidence that TLD in all cases constitutes a better classroom option than LLD. Not only might it be that this study failed to fully capture the advantages of LLD, but also it is certain that the observed benefits of TLD greatly depend on a variety of individual and contextual factors, including the skill of the instructor in managing interactions effectively and a positive class rapport. Indeed, the recordings of both TLD classes suggest that prior to the study, each instructor had established a setting in which learners felt comfortable offering unsolicited support to peers and initiating uncued requests for teacher assistance (e.g., Example 5, line j; Example 7, line l and line n).
Although such positive group dynamics were also evident in the LLD classes, they probably more greatly enhanced the TLD treatment, given the centrality of the teacher in building a sense of collaboration and support among learners during whole-group tasks. Still, the fact that the TLD group engaged in one to two additional tasks per day while the LLD group performed posttask follow-ups raises the issue of time management within the lesson as a possible factor affecting these results, in addition to the discourse format. It also remains for future research to determine whether the successful discourse management shown here for TLD
Language Learning 61:Suppl. 1, June 2011, pp. 141–188 178
Toth
Teacher- and Learner-Led Discourse
can be reproduced with learners belonging to other social, cultural, and age groups and whether similar results can be obtained for other teacher-selected objects of instruction. This will surely be a crucial aspect for future study, given the intuitive advantage of allowing learners to set their own attentional focus on the L2 during LLD and the transcript evidence here and elsewhere indicating that this is indeed what they do when working on tasks with peers. By showing stronger learning outcomes for an instructional object via task-based, teacher-led discourse, vis-` -vis learner-led, small-group discourse, a this study provides evidence against a common pedagogical belief that learner groups are necessarily a preferable format for achieving development with target L2 structures via a communicative focus on form. Instead, the data here suggest that although there are identifiable advantages to both formats, the teacher’s provision of input models, focused elicitation, procedural assistance, and form-focused feedback might have more greatly benefited whole-class learners’ performance with the instructional object than the linguistic support provided among learners in small groups. Clearly, some of the problems that have been documented for TLD stem not simply from the teacher’s greater control over discourse but rather from ineffective choices in the exercise of that control during classroom interactions. In the first Spiderman movie of the American series starring Tobey Maguire (Lee & Caracciolo, 2002), the superhero’s father warns him that “with great power comes great responsibility.” Far be it for Hollywood to have any relevance in applied linguistics, but when assessing the teacher’s potential to build a beneficial discourse for learners’ linguistic development, it is clear that so much depends on wise moment-by-moment choices and an overall technique that takes morphosyntactic complexity and discourse cohesion into account. Rather than categorically eschewing teacher-led discourse as antithetical to the principles of a “learner-centered curriculum,” L2 teacher education programs must, for the sake of learners, do a more adequate job of training teachers to effectively and responsibly conduct whole-class interactions. Nonetheless, rather than urging a pendulum swing to the opposite extreme, with TLD the hero and LLD the villain, it appears that the best approach to facilitating L2 development comes from a careful analysis of the strengths and limitations of each as a discursive pedagogical tool, and from a sound training in maximizing their respective potentials. Clearly, an optimal L2 pedagogy would call for the principled integration of both formats into classroom practice, matching instructional goals with the benefits that either format provides.
Revised version accepted 13 August 2007
179
Language Learning 61:Suppl. 1, June 2011, pp. 141–188
Toth
Teacher- and Learner-Led Discourse
Notes
1 Loschky and Bley-Vroman (1993, pp. 140–141) argue that making target structures “essential” on production tasks is difficult, given learners’ freedom to use alternate structures. This will indeed be evident in the qualitative data of the present study. 2 The larger study (Toth, 1997) also included an output-free, teacher-led “processing instruction” group, as described in Lee and VanPatten (2003), in addition to the LLD and TLD groups described here. This group’s results are reported in Toth (1997, 2000, 2003, 2006).
3 Note that Pica’s (1987) study comparing an “information exchange” activity in both a small-group and teacher-led format does not indicate how many students were involved in the whole-class interaction. There, each learner was given a unique piece of visual information about where a flower ought to be planted in a garden, and by pooling each learner’s information, a complete picture of the garden emerged. For the present study, this design was unfeasible, given the large class sizes and the fact that such complex interactional planning would have had to be sustained for a full week of instruction rather than a single instructional activity. 4 All materials given to instructors are available upon request. 5 In the Version A picture description, the middle-voice verbs were romper (to break) and mojar (to wet); in Version B, they were cocinar (to cook) and descomponer (to break down).
The passive or impersonal verbs in Version A were traer (to bring) and destruir (to destroy); in Version B, they were o´r (to hear) and ver (to see).
ı 6 A number of recent studies have incorporated a picture interpretation component within GJ tasks to ensure that learner judgments are tied to knowledge of L2 form-meaning connections (Bley-Vroman & Yoshinaga, 2000; Montrul, 1999, 2000).
Although the lack of picture interpretation on this task means that some learners could have answered without understanding an item’s meaning, the statistically reliable distinctions made by these learners between different verb classes and grammatical versus ungrammatical items, as reported in Toth (1997, 2000, 2003), suggest that such blind answering did not occur on a large scale. All test items used vocabulary prominently appearing in the curriculum prior to the test, and learners were instructed to ask for any lexical help they needed. The goal of the GJ task was thus only to have learners indicate whether sentences using familiar vocabulary were possible in Spanish with or without se. In order to reduce the impact of explicit “monitoring” on task performance, a 25-min limit was set for the entire test and learners were instructed to answer “by feel” on the GJ task. 7 In Version A of the GJ task, middle-voice verbs were cocinar, relajar, and llenar (to cook, relax, and fill); in Version B, they were: descomponer, romper, and encender (to break down, break, and turn on).
The passive or impersonal verbs in Version A were poner, comprar, ver, fumar, and comer (to put, buy, see, smoke, and eat); in Version B, they were: traer, hacer, llevar, entrar, and decidir (to bring, make, wear, enter, and decide).
Language Learning 61:Suppl. 1, June 2011, pp. 141–188
180
Toth
Teacher- and Learner-Led Discourse
8 The high standard deviation for the NS group reflects their frequent use of alternatives to se. The most common were transitive sentences for externally caused events (e.g., Ellos destruyeron la casa [They destroyed the house]) and resultant-state passives for inchoative ones (El vaso est´ roto [The glass is a broken]).
Within Pustejovsky’s (1995) analysis of transitivity, this distribution of alternatives to se is to be expected, given the varying prominence of external agents in the two verb types. Because the task was open-ended, the frequency of nontarget responses was not taken to indicate a lack of validity. Instead, high standard deviations resulted from the necessity of imposing binary scoring on a task that yielded multiple answer categories. The NS performance is indeed intended only to provide a basis for comparison with the learners. 9 The Fisher PLSD simply considers each possible group pairing, whereas the Tukey HSD does so after adjusting the threshold of significance for multiple comparisons. The rationale for using Fisher is that the Tukey adjustment is not necessary, given that the pairwise comparisons occur after reaching a significant ANOVA. Generally, the Fisher is recommended if one simply wants to know which group is more likely to have an effect, whereas the Tukey is preferred if one wants to be sure that a given group will yield an effect (Hilton & Armstrong, 2006).
10 This recording was chosen over that of the other class due principally to sound quality.
References
Adair-Hauck, B., & Donato, R. (1994).
Foreign language explanations within the Zone of Proximal Development. Canadian Modern Language Review, 50, 532–557. Ant´ n, M. (1999).
The discourse of a learner-centered classroom: Sociocultural o perspectives on teacher-learner interaction in the second-language classroom. Modern Language Journal, 83, 303–318. Austin, J. L. (1962).
How to do things with words. Oxford: Oxford University Press. Bley-Vroman, R., & Yoshinaga, N. (2000).
The acquisition of multiple wh-questions by non-native speakers of English. Second Language Research, 16, 3–26. Breen, M. P., & Candlin, C. N. (1980).
The essentials of a communicative curriculum in language teaching. Applied Linguistics, 1, 89–112. Brooks, F. B. (1993).
Some problems and caveats in “communicative” discourse: Toward a conceptualization of the foreign language classroom. Foreign Language Annals, 26, 231–242. Brooks, F. B., Donato, R., & McGlone, J. V (1997).
When are they going to say “it” . right? Understanding learner talk during pair-work activity. Foreign Language Annals, 30, 524–541. Bruhn de Garavito, J. (1999).
Adult SLA of SE constructions in Spanish. Spanish Applied Linguistics, 3, 247–296.
181 Language Learning 61:Suppl. 1, June 2011, pp. 141–188
Toth
Teacher- and Learner-Led Discourse
Buckwalter, P. (2001).
Repair sequences in Spanish L2 dyadic discourse: A descriptive study. Modern Language Journal, 85, 380–397. Carroll, S. E. (2001).
Input and evidence: The raw material of second language acquisition. Philadelphia: Benjamins. Cazden, C. B. (2001).
Classroom discourse: The language of teaching and learning (2nd ed.).
Portsmouth, NH: Heinemann. Cullen, R. (1998).
Teacher talk and the classroom context. ELT Journal, 52, 179–187. deBot, K. (1996).
The psycholinguistics of the output hypothesis. Language Learning, 46, 529–555. DeKeyser, R. (2003).
Implicit and explicit learning. In C. J. Doughty & M. H. Long (Eds.), The handbook of second language acquisition (pp. 311–348).
Oxford: Blackwell. Dobrobie-Sorin, C. (1998).
Impersonal se constructions in Romance and passivization of unergatives. Linguistic Inquiry, 29, 399–437. Donato, R. (1994).
Collective scaffolding in second language learning. In J. Lantolf & G. Appel (Eds.), Vygotskian approaches to second language learning. Norwood, NJ: Ablex. Donato, R., & Brooks, F. B. (2004).
Literary discussions and advanced speaking functions: Researching the (dis)connection. Foreign Language Annals, 37(2), 183–199. Doughty, C. (2001).
Cognitive underpinnings of focus on form. In P. Robinson (Ed.), Cognition and second language instruction (pp. 206–257).
Cambridge: Cambridge University Press. Doughty, C., & Pica, T. (1986).
“Information gap” tasks: Do they facilitate second language acquisition? TESOL Quarterly, 20, 305–325. Doughty, C., & Williams, J. (1998a).
Issues and terminology. In C. Doughty & J. Williams (Eds.), Focus on form in classroom second language acquisition (pp. 1–12).
Cambridge: Cambridge University Press. Doughty, C., & Williams, J. (Eds.).
(1998b).
Focus on form in classroom language acquisition. Cambridge: Cambridge University Press. Ellis, R. (1997).
SLA research and language teaching. Oxford: Oxford University Press. Ellis, R. (2003).
Task-based learning and teaching. Oxford: Oxford University Press. Foster, P., & Ohta, A. S. (2005).
Negotiation for meaning and peer assistance in second language classrooms. Applied Linguistics, 26, 402–430. Fotos, S. (1993).
Consciousness raising and noticing through focus on form: Grammar task performance versus formal instruction. Applied Linguistics, 14, 385–403. Fotos, S. (1994).
Integrating grammar instruction and communicative language use through grammar consciousness-raising tasks. TESOL Quarterly, 28, 323–351. Fotos, S. (2002).
Structure-based interactive tasks for the EFL grammar learner. In E. Hinkel & S. Fotos (Eds.), New perspectives on grammar teaching in second language classrooms (pp. 135–154).
Mahwah, NJ: Erlbaum.
Language Learning 61:Suppl. 1, June 2011, pp. 141–188 182
Toth
Teacher- and Learner-Led Discourse
Gass, S. (2003).
Input and interaction. In C. J. Doughty & M. H. Long (Eds.), The handbook of second language acquisition (pp. 224–256).
Malden, MA: Blackwell. Hall, J. K. (1995).
“Aw man, where you goin’?”: Interaction and the development of L2 interactional competence. Issues in Applied Linguistics, 6, 37–62. Hall, J. K. (2004).
“Practicing speaking” in Spanish: Lessons from a high school foreign language classroom. In D. Boxer & A. D. Cohen (Eds.), Studying speaking to inform second language learning (pp. 68–87).
Clevedon, UK: Multilingual Matters. Harrington, M. (2001).
Sentence processing. In P. Robinson (Ed.), Cognition and second language instruction (pp. 91–124).
Cambridge: Cambridge University Press. Hilton, A., & Armstrong, R. (2006, September).
Stat note 6: Post hoc ANOVA tests. Microbiologist, 7, 34–36. Hymes, D. (1972).
On communicative competence. In J. B. Pride & J. Holmes (Eds.), Sociolinguistics. Harmondsworth, UK: Penguin Books. Izumi, S. (2003).
Comprehension and production processes in second language learning: In search of the psycholinguistic rationale of the output hypothesis. Applied Linguistics, 24, 168–196. Jacobs, G. M. (1998).
Cooperative learning or just grouping students: The difference makes a difference. In W. A. Renandya & G. M. Jacobs (Eds.), Learners and language learning (pp. 145–171).
Singapore: SEAMEO. Johnson, K. (1996).
Language teaching and skill learning. Oxford: Blackwell. Johnson, K. E. (1995).
Understanding communication in second language classrooms. Cambridge: Cambridge University Press. Juffs, A. (1998).
Main verb vs. reduced relative clause ambiguity resolution in second language sentence processing. Language Learning, 48, 107–147. Juffs, A. (2004).
Representation, processing, and working memory in a second language. Transactions of the Philological Society, 102, 199–225. Juffs, A., & Harrington, M. (1995).
Parsing effects in second language sentence processing: Subject and object asymmetries in Wh-extraction. Studies in Second Language Acquisition, 17, 483–516. Kormos, J. (1999).
Monitoring and self-repair in L2. Language Learning, 49, 303–342. Krashen, S., Sferlazza, V Feldman, L., & Fathman, A. (1976).
Adult performance on ., the SLOPE test: More evidence for a natural sequence in adult second language acquisition. Language Learning, 26, 145–151. Larsen-Freeman, D. (1975).
The acquisition of grammatical morphemes by adult ESL students. TESOL Quarterly, 9, 409–430. Lee, J. F. (2000).
Tasks and communicating in language classrooms. New York: McGraw-Hill. Lee, J. F., & VanPatten, B. (2003).
Making communicative language teaching happen (2nd ed.).
New York: McGraw-Hill. Lee, S., & Caracciolo, J. M. (Producers), D. Koepp (Writer), & S. Raimi (Director).
(2002).
Spider-Man [Motion picture]. United States: Columbia Pictures.
183
Language Learning 61:Suppl. 1, June 2011, pp. 141–188
Toth
Teacher- and Learner-Led Discourse
Leeman, J. (2003).
Recasts and second language development: Beyond negative evidence. Studies in Second Language Acquisition, 25, 37–64. Leemann-Guthrie, E. (1984).
Intake, communication, and second-language teaching. In S. Savignon & M. Berns (Eds.), Initiatives in communicative language teaching: A book of readings (pp. 35–54).
Reading, MA: Addison-Wesley. Levelt, W. J. M. (1989).
Speaking: From intention to articulation. Cambridge, MA: MIT Press. Levin, B., & Rappaport-Hovav, M. (1995).
Unaccusativity: At the syntax-lexical semantics interface. Cambridge, MA: MIT Press. Long, M. H. (1981).
Input, interaction and second language acquisition. In H. Winitz (Ed.), Native language and foreign language acquisition. Annals of the New York Academy of Sciences: Vol. 379 (pp. 259–278).
New York: New York Academy of Sciences. Long, M. H. (1985).
A role for instruction in second language acquisition: Task-based language teaching. In K. Hyltenstam & M. Pienemann (Eds.), Modeling and assessing second language acquisition (pp. 77–99).
Clevedon, UK: Multilingual Matters. Long, M. H. (1991).
Focus on form: A design feature in language teaching methodology. In K. deBot, D. Coste, R. Ginsberg, & C. Kramsch (Eds.), Foreign language research in cross-cultural perspectives (pp. 39–52).
Amsterdam: Benjamins. Long, M. H. (1996).
The role of the linguistic environment in second language acquisition. In W. C. Ritchie & T. K. Bhatia (Eds.), Handbook of second language acquisition (pp. 413–468).
New York: Academic Press. Long, M. H., & Crookes, G. (1993).
Units of analysis in syllabus design: The case for task. In G. Crookes & S. Gass (Eds.), Tasks in a pedagogical context: Integrating theory and practice (pp. 9–54).
Clevedon, UK: Multilingual Matters. Long, M. H., & Porter, P. A. (1985).
Group work, interlanguage talk, and second language acquisition. TESOL Quarterly, 19, 207–228. Loschky, L., & Bley-Vroman, R. (1993).
Grammar and task-based methodology. In G. Crookes & S. Gass (Eds.), Tasks and language learning: Integrating theory and practice (pp. 123–163).
Clevedon, UK: Multilingual Matters. Lyster, R. (2001).
Negotiation of form, recasts, and explicit correction in relation to error types and learner repair in immersion classrooms. In R. Ellis (Ed.), Form-focused instruction and second language learning (Vol. 51, pp. 265–301).
Malden, MA: Blackwell. Mackey, A. (1999).
Input, interaction, and second language development. Studies in Second Language Acquisition, 21, 557–587. McCormick, D. E., & Donato, R. (2000).
Teacher questions as scaffolded assistance in an ESL classroom. In J. K. Hall & L. S. Verplaetse (Eds.), Second and foreign language learning through classroom interaction (pp. 183–202).
Mahwah, NJ: Erlbaum.
Language Learning 61:Suppl. 1, June 2011, pp. 141–188 184
Toth
Teacher- and Learner-Led Discourse
Mehan, H. (1979).
Learning lessons: Social organization in the classroom. Cambridge, MA: Harvard University Press. Montrul, S. (1999).
Causative errors with unaccusative verbs in L2 Spanish. Second Language Research, 15, 191–219. Montrul, S. (2000).
Causative psych verbs in Spanish L2 acquisition. In R. Leow & C. Sanz (Eds.), Spanish applied linguistics at the turn of the millennium (pp. 97–118).
Somerville, MA: Cascadilla Press. Montrul, S. (2004).
The acquisition of Spanish: Morphosyntactic development in monolingual and bilingual L1 acquisition and adult L2 acquisition. Philadelphia: Benjamins. Morris, F. A. (2002).
Negotiation moves and recasts in relation to error types and learner repair in the foreign language classroom. Foreign Language Annals, 35, 395–404. Musumeci, D. (1996).
Teacher-learner negotiation in content-based instruction: Communication at cross-purposes? Applied Linguistics, 17, 286–325. Myles, F., Mitchell, R., & Hooper, J. (1999).
Interrogative chunks in French L2: A basis for creative construction? Studies in Second Language Acquisition, 21, 49–80. National Standards in Foreign Language Learning Project. (1999).
Standards for foreign language learning in the 21st century including Chinese, Classical Languages, French, German, Italian, Japanese, Portuguese, Russian, and Spanish. Lawrence, KS: Allen Press. Norris, J. M., & Ortega, L. (2000).
Effectiveness of L2 instruction: A research synthesis and quantitative meta-analysis. Language Learning, 50, 417–528. Nunan, D. (1987).
Communicative language teaching: Making it work. ELT Journal, 41, 136–145. Nunan, D. (1989).
Designing tasks for the communicative classroom. Cambridge: Cambridge University Press. Nunan, D. (1990).
The questions teachers ask. JALT Journal, 12, 187–202. Ohta, A. S. (2000).
Rethinking recasts: A learner-centered examination of corrective feedback in the Japanese language classroom. In J. K. Hall & L. S. Verplaetse (Eds.), Second and foreign language learning through classroom interaction (pp. 47–72).
Mahwah, NJ: Erlbaum. Ohta, A. S. (2001).
Second language acquisition processes in the classroom: Learning Japanese. Mahwah, NJ: Erlbaum. Paulston, C. B. (1972).
The sequencing of structural pattern drills. TESOL Quarterly, 5, 197–208. Pica, T. (1987).
Second-language acquisition, social interaction, and the classroom. Applied Linguistics, 8, 3–21. Pica, T. (1994).
Research on negotiation: What does it reveal about second language learning, processes and outcomes? Language Learning, 44, 493–527. Pica, T. (2005).
Classroom learning, teaching, and research: A task-based perspective. Modern Language Journal, 89, 339–352.
185 Language Learning 61:Suppl. 1, June 2011, pp. 141–188
Toth
Teacher- and Learner-Led Discourse
Pica, T., & Doughty, C. (1985).
Input and interaction in the communicative language classroom: A comparison of teacher-fronted and group activities. In S. Gass & C. Madden (Eds.), Input in second language acquisition (pp. 115–132).
Cambridge, MA: Newbury House. Pica, T., Holliday, L., Lewis, N., & Morgenthaler, L. (1989).
Comprehensible output as an outcome of linguistic demands on the learner. Studies in Second Language Acquisition, 11, 63–90. Pica, T., Kanagy, R., & Falodun, J. (1993).
Choosing and using communication tasks for second language acquisition. In G. Crookes & S. Gass (Eds.), Tasks and language learning: Integrating theory and practice (pp. 9–34).
Clevedon, UK: Multilingual Matters. Pica, T., Kang, H.-S., & Sauro, S. (2006).
Information gap tasks: Their multiple roles and contributions to interaction research methodology. Studies in Second Language Acquisition, 28, 301–338. Porter, P. A. (1986).
How learners talk to each other: Input and interaction in task-centered discussions. In R. R. Day (Ed.), Talking to learn: Conversation in second language acquisition (pp. 200–222).
Rowley, MA: Newbury House. Prabhu, N. S. (1987).
Second language pedagogy. Oxford: Oxford University Press. Pritchett, B. L. (1992).
Grammatical competence and parsing performance. Chicago: University of Chicago Press. Pustejovsky, J. (1995).
The generative lexicon. Cambridge, MA: MIT Press. Raposo, E., & Uriagereka, J. (1996).
Indefinite SE. Natural Language and Linguistic Theory, 14, 749–810. Samuda, V (2001).
Guiding relationships between form and meaning during task . performance: The role of the teacher. In M. Bygate, P. Skehan, & M. Swain (Eds.), Researching pedagogic tasks: Second language learning, teaching, and testing (pp. 119–140).
Harlow, UK: Longman. Samuda, V & Bygate, M. (2008).
Tasks in second language learning. Basingstoke, ., UK: Palgrave Macmillan. Schmidt, R. W. (2001).
Attention. In P. Robinson (Ed.), Cognition and second language instruction (pp. 3–32).
Cambridge: Cambridge University Press. Scovel, T. (1998).
Psycholinguistics. Oxford: Oxford University Press. Searle, J. (1969).
Speech acts. Cambridge: Cambridge University Press. Seedhouse, P. (1999).
Task-based interaction. ELT Journal, 53, 149–156. Seedhouse, P. (2004).
The interactional architecture of the language classroom: A conversation analysis perspective. Malden, MA: Blackwell. Sharwood-Smith, M. (2004).
In two minds about grammar: On the interaction of linguistic and metalinguistic knowledge in performance. Transactions of the Philological Society, 102, 255–280. Sinclair, J. M., & Coulthard, R. M. (1975).
Toward an analysis of discourse: The English used by teachers and pupils. London: Oxford University Press.
Language Learning 61:Suppl. 1, June 2011, pp. 141–188
186
Toth
Teacher- and Learner-Led Discourse
Skehan, P. (1996).
A framework for the implementation of task-based instruction. Applied Linguistics, 17, 38–62. Skehan, P. (1998).
A cognitive approach to language learning. Oxford: Oxford University Press. Skehan, P., & Foster, P. (1999).
The influence of task structure and processing conditions on narrative retellings. Language Learning, 49, 93–120. Swain, M. (1985).
Communicative competence: Some roles of comprehensible input and comprehensible output in its development. In S. Gass & C. Madden (Eds.), Input in second language acquisition. Cambridge, MA: Newbury House. Swain, M. (1995).
Three functions of output in second language learning. In G. Cook & B. Seidlhofer (Eds.), Principle and practice in applied linguistics: Studies in honour of H. G. Widdowson (pp. 125–144).
Oxford: Oxford University Press. Swain, M. (1998).
Focus on form through conscious reflection. In C. Doughty & J. Williams (Eds.), Focus on form in classroom second language acquisition (pp. 64–82).
Cambridge: Cambridge University Press. Swain, M. (2000).
The output hypothesis and beyond: Mediating acquisition through collaborative dialogue. In J. Lantolf (Ed.), Sociocultural theory and second language learning (pp. 97–114).
Oxford: Oxford University Press. Swain, M., & Lapkin, S. (1995).
Problems in output and the cognitive processes they generate: A step towards second language learning. Applied Linguistics, 16, 371–391. Swain, M., & Lapkin, S. (1998).
Interaction and second language learning: Two adolescent French immersion students working together. Modern Language Journal, 82, 320–337. Swain, M., & Lapkin, S. (2001).
Focus on form through collaborative dialogue: Exploring task effects. In M. Bygate, P. Skehan, & M. Swain (Eds.), Researching pedagogic tasks: Second language learning, teaching, and testing (pp. 99–118).
Harlow, UK: Longman. Terrell, T. (1991).
The role of grammar instruction in a communicative approach. Modern Language Journal, 75, 52–63. Terrell, T., Andrade, M., Egasse, J., & Mu˜ oz, E. M. (1994).
Dos Mundos (3rd ed.).
n New York: McGraw-Hill. Tharp, R. G., & Gallimore, R. (1988).
Rousing minds to life: Teaching, learning, and schooling in social context. Cambridge: Cambridge University Press. Toth, P. D. (1997).
Linguistic and pedagogical perspectives on acquiring second-language morpho-syntax: A look at Spanish se. Unpublished doctoral dissertation, University of Pittsburgh, Pittsburgh. Toth, P. D. (2000).
The interaction of instruction and learner-internal factors in the acquisition of L2 morphosyntax. Studies in Second Language Acquisition, 22, 169–208. Toth, P. D. (2003).
Psych verbs and morphosyntactic development in instructed L2 Spanish. In S. Montrul & F. Ord´ nez (Eds.), Linguistic theory and language o˜
187 Language Learning 61:Suppl. 1, June 2011, pp. 141–188
Toth
Teacher- and Learner-Led Discourse
development in Hispanic languages: Papers from the 5th Hispanic linguistics symposium and the 4th conference on the acquisition of Spanish and Portuguese (pp. 468–497).
Somerville, MA: Cascadilla Press. Toth, P. D. (2004).
When grammar instruction undermines cohesion in L2 Spanish classroom discourse. Modern Language Journal, 88, 14–30. Toth, P. D. (2006).
Processing Instruction and a role for output in second language acquisition. Language Learning, 56, 319–385. Van den Branden, K. (1997).
Effects of negotiation on language learners’ output. Language Learning, 47, 589–636. Van den Branden, K. (2000).
Does negotiation of meaning promote reading comprehension? A study of multilingual primary school classes. Reading Research Quarterly, 35, 426–443. Van Den Branden, K. (Ed.).
(2006).
Task-based language education: From theory to practice. Cambridge: Cambridge University Press. van Lier, L. (1996).
Interaction in the language curriculum: Awareness, autonomy, and authenticity. London: Longman. VanPatten, B. (1990).
Attending to form and content in the input. Studies in Second Language Acquisition, 12, 287–301. VanPatten, B. (1996).
Input processing and grammar instruction in second language acquisition: Theory and research. Norwood, NJ: Ablex. VanPatten, B. (2004a).
Input and output in establishing form-meaning connections. In B. VanPatten, J. Williams, S. Rott, & M. Overstreet (Eds.), Form-meaning connections in second language acquisition (pp. 29–47).
Mahwah, NJ: Erlbaum. VanPatten, B. (Ed.).
(2004b).
Processing Instruction: Theory, research, and commentary. Mahwah, NJ: Erlbaum. Varonis, E. M., & Gass, S. (1985).
Non-native/non-native conversations: A model for negotiation of meaning. Applied Linguistics, 6, 71–90. Wells, G. (1993).
Reevaluating the IRF sequence: A proposal for the articulation of theories of activity and discourse for the analysis of teaching and learning in the classroom. Linguistics and Education, 5, 1–37. Wells, G. (1998).
Some questions about direct instruction: Why? To whom? How? and When? Language Arts, 76(1), 27–35. White, L. (2003).
Second language acquisition and Universal Grammar. Cambridge: Cambridge University Press. Williams, J. (1999).
Learner-generated attention to form. Language Learning, 49, 583–625. Wood, D., Bruner, J. S., & Ross, G. (1976).
The role of tutoring in problem solving. The Journal of Child Psychology and Psychiatry, 17, 89–100.
Language Learning 61:Suppl. 1, June 2011, pp. 141–188
188