Native (NS) and Nonnative Speaker (NNS) Models for the Teaching of Speaking

A number of recent publications debate the issue of NS versus learner and NNS corpora (Flowerdew, 2000; Nero, 2000; Prodromou, 2003; Seidlhofer, 2001, this volume; Warschauer, 2000). Prodromou (1997), whose work is based on a mixed NS/NNS spoken English corpus, had already raised issues concerning the undermining effect of spoken corpora for NNS faced with varieties and cultures that they can “never master, never own” (p. 5). Reacting to similar concerns, Seidlhofer proposes a spoken corpus of English as a lingua franca (ELF), which will help to profile ELF as robust and independent of English as a native language and may establish “something like an index of communicative redundancy” with pedagogical applications (Seidlhofer, 2001, p. 147). The shift away from the NS as the sole model for spoken pedagogy is further underlined by the introduction into the debate of terms aimed at leveling the playing field between NS and NNS as potential models. Building on earlier work by Leung, Harris, and Rumpton (1997) such terms include “expert users” (McCarthy, 2001, pp. 139–142) and “successful users of English” (SUEs) (Prodromou, 2003), with a focus on the modeling of successful language users (whether NS or NNS) in nonpedagogical contexts. Meanwhile the The Louvain International Database of Spoken English Interlanguage (LINDSEI), set up in 1995 (see De Cock, 1998, 2000), provides spoken data for the analysis of learner language (see also Granger, Hung, & Petch-Tyson, 2002).

From “Spoken Language” to “Speaking:” Relevant Data for Input

Bygate (1998) observed that second language speaking (except for pronunciation) had been something of a Cinderella compared with listening, reading, and writing, but recent research is redressing that imbalance (see also Bygate, 2001). Current debates include whether speaking materials do or should truly reflect naturally occurring spoken language, what teachers' and learners' perceptions are of the importance of using real speech samples (Hughes, 2002; Timmis, 2002), and whether learner output reflects authentic spoken patterns. However, there is by no means universal agreement on the relationship between descriptions of spoken language and methods, materials, and activities.

Naturally occurring data have not been the only source for the pedagogic modeling of speech. Debate continues as to how elicited speech data compare with naturally occurring speech, especially data generated by discourse completion tasks (DCTs), which aim to describe what constitutes spoken pragmatic competence based on informants' intuitive reactions to situational prompts. Golato (2003) looks at differences between naturally occurring data analyzed using conversation analysis (CA), and DCTs constructed on the basis of naturally occurring conversations. Golato notes that while CA clearly enables the study of language organization in natural settings, CA analyses cannot necessarily be extrapolated to other situations. DCTs, however, may be construed as a condensation of the informant's prior experience with language. Golato also evaluates other data collection procedures (questionnaires, recall protocols, role play, etc.) and discusses the problems and prospects for generalizing from different kinds of data. Billmyer and Varghese (2000) argue that, in practice, DCTs will continue to be an important resource for spoken pragmatic pedagogy. Their study examines systematic modifications (adding information on social and contextual variables) to DCT situational prompts and responses from NS and NNS. Modified prompts produced longer, more elaborated responses from both NS and NNS informants. Yuan (2001) further finds that orally produced DCTs generate more natural speech features than written DCTs. Such findings suggest that carefully constructed DCTs are not without positive value in speaking pedagogy.

McCarthy (2001) and McCarthy and Carter (2001) have continued to argue for a description of spoken grammar and vocabulary that is independent, based on the evidence of spoken corpora, and not presented merely as a list of deviations from written norms, as a means of providing systematic linguistic input for pedagogies of speaking (see also Hughes, 2002; Carter, Hughes, & McCarthy, 2000). However, not least of the problems of using spoken data in pedagogy is the question of what the basic unit of spoken grammar might consist of. In the absence of well-formed sentences (which are famously rare in casual, NS conversation), clause- and phrasebased units seem to be more appropriate (Burns, 2001; Foster, Tonkyn, & Wigglesworth, 2000). On the other hand, corpus analysts and other linguists alike have in recent years stressed the central role of fixed “chunks” of various kinds in everyday, fluent speech. Wray (2000, 2002), investigating formulaic sequences (which include idioms, collocations, and institutionalized sentence frames), stresses that such sequences circumvent the analytical processes associated with the interpretation of open syntactic frames in terms of both reception and production. Attempts to encourage the analysis of formulaic sequences in second language pedagogy are criticized as “pursuing native-like linguistic usage by promoting entirely unnative-like processing behaviour” (Wray, 2000, p. 463). Wray's work attempts to move away from a static, behaviorist account of formulaic sequences, emphasizing their nature as “a dynamic response to the demands of language use,” which “will manifest differently as those demands vary from moment to moment and speaker to speaker” (Wray, 2002, p. 5). Hunston (2002), discussing lexicogrammatical patterns as evidenced in the Bank of English corpus, suggests that sequences such as verb complementation patterns, if learned and stored as holistic units, can contribute greatly to spoken fluency, producing, when chained together in connected, meaningful strings, “pattern flow.” Hunston advocates a task-based approach to this aspect of fluency, rejecting presentational approaches as inappropriate to the need to raise awareness of features of speaking as a whole, as opposed to learning individual patterns.

Discourse Analysis: Exchange Structure and Classroom Speaking

Kasper begins by considering the widely held view that “the IRF routine is an unproductive interactional format for the learning of pragmatics and discourse” (2001, p. 518). Exchanges consisting of the familiar classroom pattern of initiation (I), response (R), and follow-up (F), often referred to as IRF exchanges, fail to give opportunities for tackling the complex demands of everyday conversation, especially since teachers usually exercise the follow-up role, while learners often remain in passive, respondent roles. However, Kasper goes on to argue that the negative reputation enjoyed by the IRF exchange may not be the whole story and that what really matters is the kind of interactional status assigned by the teacher to individual learners: where students are seen as primary interactants in speaking activities, teachers offer them more participation rights in the conversation. Kasper refers to evidence that suggests that teachers can help their learners to become actively involved in interaction, even within the typical IRF patterning found in teacher-fronted classroom dialogue.

McCarthy (2002, 2003) argues that responding and follow-up moves play a key role in “listenership,” the manifestation of engagement in the discourse even when one is not in the role of main speaker, a situation NNSs often find themselves in. For McCarthy, listenership is not the same as “listening” in the conventional four skills paradigm, and is an important component of the speaking skill. However, Ohta (2001) finds that the overwhelming majority of classroom follow-up moves are spoken by the teacher; learners get few opportunities to use typical listener followups and only experience the teacher's moves as peripheral participants. Ohta argues for peer-to-peer interaction as providing the best opportunities for learners to produce appropriate listener behavior. Notwithstanding, exposure to the teacher's use of follow-up moves along with explicit guidance on the use of responsive moves can help students gradually move toward more productive use in peer-to-peer speaking activities. S. Walsh (2002) distinguishes different modes of teacher talk and illustrates how these may hinder or optimize learner contributions. More generally, Hall and M. Walsh (2002) give a relevant and wide-ranging survey of current research into teacher–student interaction and language learning (see also Morita, 2000).

Conversation Analysis and Speaking Tasks

Advocacy of conversation analysis (CA) as a means of understanding and improving speaking in pedagogical contexts has continued to grow in recent years. Wong (2000) notes that CA illuminates how local choices unfold in interaction and can home in on aspects of talk which are relevant for the participants themselves. Ducharme and Bernard (2001) argue along the same lines in their study of learners of French, using micro-analyses of videotaped interactions and retrospective interviews to gain insights into the perspectives of participants. Mori (2002) uses CA to analyze a speaking activity in a class of NNS learners of Japanese, where students exchanged experiences and opinions with Japanese NSs invited to the class. The resulting interaction resembled an interview, with a succession of questions by the students and answers from the NS guests. Interestingly, more natural discussion came about when students made spontaneous utterances and when they seemed to be attending more to the moment-by-moment unfolding of the talk. Mori ponders how the talk revealed the speakers' orientation toward the institutionalized nature of the task. Overall, the CA argument is that factors of task design and execution can influence the resultant talk in ways that CA can make plain, with clear implications for the improvement of the design and implementation of speaking tasks. However, Rampton, Roberts, Leung, and Harris (2002) warn of the lack of a “learning” dimension in CA studies of this kind. Because CA is a very local kind of analysis, it lends itself less easily to providing evidence of actual development of speaking ability over time.

Cognitive Approaches to Speaking Tasks

Other recent work examines the design and implementation of speaking tasks from the point of view of fluency, complexity and accuracy of production (see Bygate, 2001 for an overview of the evolution of research in this area). Robinson (2001) claims that increasing the cognitive complexity of speaking tasks affects production, with greater lexical variation manifested in more complex versions and greater fluency evidenced in simpler versions of the task. Yuan and Ellis (2003) assert that pretask planning positively aids learners' spoken production, especially with regard to fluency and complexity, albeit accuracy may not benefit so obviously. Yuan and Ellis also examine online planning, where learners are given unlimited time to formulate and monitor their speech while performing, and claim that online planning positively influences accuracy and grammatical complexity. Repetition and recycling and their contribution to the increasing integration over time of fluency, complexity, and accuracy of oral production is also receiving attention (Bygate, 2001; Lynch & Maclean, 2001). Additionally, the role of the teacher vis-à-vis the design and execution of tasks and the teacher's role of providing scaffolding (see the section on Vygotsky, below) to help learners develop their oral competence has come under the spotlight (Samuda, 2001).

The Vygotskian Perspective on Speaking

Since the publication of Lantolf and Appel's (1998) and Lantolf's (2000) influential works on Vygotsky, interest has grown in how such a perspective feeds into speaking pedagogy. Of central importance are the notions of scaffolding and the zone of proximal development (ZPD). Scaffolding is the cognitive support provided by an adult or other guiding person to aid a child or learner, and is realized in dialogue so that the child/learner can come to make sense of difficult tasks. The ZPD is the distance between where the child/learner is developmentally and what s/he can potentially achieve in interaction with adults or more capable peers (Vygotsky, 1978). In the Vygotskian paradigm, instructors and pupils (or peers) interactively coconstruct the arena for development; it is not predetermined and has no lockstep limits or ceiling. Meaning is created in dialogue (including dialogue with the self, often manifested in “private speech”) during goal-directed activities. Attempts to see how these notions operate in reality in classrooms include Machado (2000), who demonstrates how peer-to-peer scaffolding in the preparatory phases of spoken classroom tasks (mutual help with the interpretation of the tasks and the wording of meanings) is reflected in evidence of internalization of such help in the performance phases of the same tasks. Machado suggests that peer-to-peer scaffolding may be just as important as expert–novice scaffolding, a theme reiterated by Ko, Schallert, and Walters (2003), who examine what marks out higher quality from lower quality negotiation-of-meaning interactions during a storytelling task (see also Kasper, 2001). As well as through teachers' contributions, improved storytelling was assisted by peer contributions in a negotiation phase between the first and second telling of a story, which, the researchers suggest, shows learners playing a central role in scaffolding (see also Shumin, 2002).

However, Kinginger (2002) warns against the incorporation of Vygotskian notions such as scaffolding and the ZPD into existing practices in oral pedagogy in ways that the notions simply become a justification for current practices (e.g., the input–output hypotheses, all and any types of pair- and group-work tasks, teacher feedback moves, etc.) rather than a genuine reexamination of the role of social interaction in language development. In this respect, Kinginger highlights the work of Merrill Swain and her associates (Swain, 2000; Nassaji & Swain, 2000; Swain & Lapkin, 2000), who see the ZPD more as an open-ended opportunity for unanticipated development and unpredictability rather than some fixed territory circumscribed by input and output or the closed cycles of teacher–learner exchanges.

Selecting Spoken Texts for Classroom Use

Burns, Gollin, and Joyce (1997) suggest that if learners are involved in spoken text analysis, it should be directly applicable to the speaking task. They suggest as an example that learners could be presented with a partially transcribed text and be given the task of listening to the tape recording so as to fill in features such as backchannel responses raise awareness of such conversational features. They also suggest that learners transcribe small amounts of their own conversation and compare particular features such as length of turn and overlaps with NS transcripts. For a teacher wishing to use spoken data in the classroom, they suggest the following analytical framework: (1) transcribe the recording, (2) give the background information to the text, (3) analyze the text using an analytical approach (e.g. SFL, IRF, etc.; see the review of Burns, 2001, above), and (4) identify the significant teaching points which arise from the analysis. This approach, as in Burns (2001), argues for teachers to be well versed in methods of analysis of spoken language and aligns with the general move toward the exploitation of attested spoken data.

Selecting Aspects of Spoken Language to Focus on in Teaching

Burns, Joyce, and Gollin (2001) offer a practical handbook for teaching spoken discourse that provides guidelines for collecting, transcribing, and using authentic language, as well as tasks and case studies. Riggenbach (1999) provides numerous template activities to practice macro- (e.g., turn-taking) and micro- (e.g., pronunciation) speaking skills. In what still remains one of the few papers to give practical direction as to syllabus design for teaching conversation, Dörnyei and Thurrell (1994) noted that teachers are often unsure about which topic areas they should focus on in teaching conversation. Based on a synthesis of research from fields such as discourse analysis, CA, and the study of communicative competence, they suggest teaching points under four headings: conversational rules and structure (e.g., openings, topic shifting and closing), conversational strategies (e.g., paraphrase, asking for repetition and clarification), functions and meaning in conversation (e.g., illocutionary functions), and social and cultural contexts (e.g., social norms of appropriateness). Boyer's (2003) textbook integrates traditional speaking domains such as pronunciation, stress patterns, intonation and language functions (e.g., making suggestions), with lexicogrammar in spoken language (e.g., question tags, pronouns), conversational strategies (changing the topic), discourselevel features (e.g., ellipsis, small talk) and situational contexts (e.g., stages in a medical consultation). Other recent textbooks on teaching conversation include Zelman and Moran (1996) and Measday (1998).

Small Talk

Shumin (2002) stresses the need for learners to be able to engage in small talk in the target language (i.e., inconsequential talk about the weather, traffic, and so on). Such interactional talk functions to create a sense of social communion. Shumin therefore suggests that from the outset adult learners should develop skill in short, interactional exchanges in which they are required to make only one or two utterances at a time (e.g., I hate rush-hour traffic — me too).

Cunningham Florez (1999) suggests that speaking lessons can follow a classic pattern of preparation, presentation, practice, evaluation, and extension, and gives the example of teaching small talk. In the preparation phase, learners are shown visuals of people in informal settings and brainstorm on what they might be saying. The presentation phase then has video clips of people doing small talk, during which students focus on the topics of talk and the language involved. The practice phase maintains the dual focus on topics and language with simple dialogues generated by the learners based on the material of the earlier stages. In the evaluation phase, learners compare their dialogues with a teacher-prepared dialogue, and discuss similarities and differences. Finally, the extension phase sends the learners out into the community (where this is feasible) to observe small talk in real settings and then report back to their class. A combination of traditional pedagogic structuring and the learner-as-researcher is at the center of this kind of activity, especially in relation to a conversational genre difficult to recreate naturally in an institutional setting.

Discussions

Green, Christopher, and Lam (2002) note that discussion skills are much neglected in the EFL/ESL classroom, and are critical of overly structured discussion activities. However, Shumin (2002) notes that the totally unstructured alternative of simply assigning a topic to students for discussion is not enough to ensure that speaking skills will be developed. In this respect, Green et al. (2002) suggest a threestage format for the implementation of successful classroom discussions that are not overly controlled: (1) prediscussion (discussion groups are formed, with four per group as the optimum, possible topics are chosen; responsibility for researching the topic are divided within the groups); (2) discussion (the groups discuss the topic while partner groups observe and monitor); and (3) postdiscussion (peer-feedback from the observer-evaluators, plus teacher feedback, and finally groups decide on ways to extend the topic or choose a new topic). Lam and Wong (2000) identify key strategies that students need in order to play an effective part in discussion: seeking clarification, clarifying oneself, and checking that other people have understood one's message. While Lam and Wong's study underscores the value of strategy training, they raise two issues: (1) the necessity to support strategy training with linguistic scaffolding and (2) the importance of peer help and cooperation in facilitating strategy use. A number of recent classroom resource books on discussions are now widely available, for example, Barnard (1997), Folse and Ivone (2002a, 2002b), Kehe (1998), and Wallwork (1997a, 1997b).

Narratives

Slade (1997), in her research into workplace talk, found that social conversations were dominated by narrative genres, but noted that such genres are rarely represented in language teaching materials. She identifies four types of stories common in casual talk (narratives, anecdotes, exempla, and recounts) and argues strongly that the generic structure of narratives can be taught explicitly and that those aspects of storytelling that are culturally specific can be discussed (for further description of everyday storytelling see Norrick, 2000). Jones (2002) proposes a strategy for teaching the narrative pattern of “the reminiscence story.” Based on Deacon (2000), Jones looks at the technique of the “split story,” which involves telling students a story, but stopping at a crucial point and inviting students to provide their own imaginative ending (see also Jones, 2001). According to Wajnryb (2003), stories are a familiar and reassuring way of acquiring language and therefore can be easily applied to the language classroom. Wajnryb's (2003) textbook aims to promote the exploration of storytelling as discourse in the language classroom and provides examples of meaningful activities which can be used in a range of teaching situations (see also Paran and Watts, 2003, for a collection of stories and related classroom activtities).

Future Directions

Hughes (2002) repeatedly stresses the need for proper social and cultural contextualization of speaking activities, and there certainly seems to be a growing awareness that speaking activities, however cleverly designed, should not take place in a void, separated from the social and cultural life of the learner; undoubtedly this will be an important direction for future research in the teaching of speaking. Alongside this, new definitions of “literacy” are emerging that are no longer singularly focused on writing, but see literacy as encompassing an awareness of the nature of written and spoken texts and the development of literacy as intimately bound up with communicative activity in the classroom (Kern, 2000; Warschauer & Whittaker, 2001).

Technology and the blurring of the traditional speech–writing division as new modes of communication emerge are also at the center of discussions of how to enhance speaking pedagogy. Payne and Whitney (2002), for example, already claim that synchronous computer-mediated communication (CMC) can indirectly improve L2 speaking proficiency by fostering the same cognitive mechanisms that underlie spontaneous conversational speech. Their study shows that an experimental group for whom two of four contact hours per week took place in an internet chatroom environment performed better in speaking than a control group. One important implication of the use of technological support, Payne and Whitney claim, is that learning environments can, by design, reduce the cognitive burden and thereby have a facilitating effect for oral production. Simpson (2002), also investigating synchronous CMC, suggests that the communication, although written, is “conversation-like” and asserts that CMC generates communicative strategies that are medium-appropriate, borrowing from conversation, but taking on a special identity of their own. Although in its early days, such research points to fresh possibilities for enhancing the teaching of speaking.

When the time comes for the next ARAL review of teaching speaking, we can expect further progress in the application of spoken corpus research. Published classroom materials are likely to increasingly reflect the growing body of research into spoken discourse, and useful insights may be available as to the global use of spoken English with implications for teaching it in an international context.