2.1 Corpus analysis

The ‘corpus revolution’ (Rundell & Stock Reference Rundell and Stock1992) has had a major and lasting impact on language learning materials in some respects. The first corpus-based learner dictionary, the Collins Cobuild English language dictionary, appeared on the market in 1987 as a result of the pioneering work of John Sinclair at the University of Birmingham. At the time it was considered by many to be over-elaborate and expensive and was only a modest commercial success (D. Willis, personal communication), but has since been widely imitated by all the major publishers, such as Oxford, Cambridge and Macmillan. Reference grammars, such as Collins Cobuild English grammar and Cambridge grammar of English, are also now corpus-informed and provide more accurate descriptions of everyday usage, as well as clarifying distinctions between spoken and written modes.Footnote ⁴ The fact that these kinds of resources are designed around word frequency lists or collocation data from mega corpora, and provide authentic samples of the language, ensures a higher level of reliability and validity absent from traditional publications, which depended more on lexicographers’ or grammarians’ intuitions and the accumulated descriptions from earlier works.

The Cobuild corpus also inspired innovative approaches to textbook design with the Collins Cobuild English course (Willis & Willis Reference Willis and Willis1988), a task-based lexical syllabus built around the most frequent 2,500 words from the corpus, which constitute 80% of all spoken and written English. Its design, unfortunately, proved to be too radical for the English language teaching community and it was not a commercial success, although it did influence later publications such as Cutting edge (Cunningham & Moor Reference Cunningham and Moor1998) and Inside out (Kay, Jones & Hird Reference Kay, Jones and Hird2001) (D. Willis, personal communication). The potential benefits of corpora on materials design can be seen very clearly in the business English textbook, Business advantage (Upper-intermediate) (Handford et al. Reference Handford, Lisboa, Koester and Pitt2011), which uses data from the Cambridge Business Corpus and the Cambridge and Nottingham Business English Corpus to both identify key discourse features of the genre and to provide semi-authentic models of spoken interaction from business contexts. For example, the listening exercise 2.14 in unit 10 is adapted from an authentic internal meeting within a luxury hotel chain, taken from CANBEC (see Handford Reference Handford2010: 44,45 for the original transcript). More than 40 minor changes were made to the original transcript in the editing process,Footnote ⁵ illustrating some of the difficulties involved in tailoring real interaction to the classroom, but the final product does, at least, recognize the value of authentic discourse for language learners. The material allows the authors to highlight persuasion strategies employed by the manager (contrast structures and their prosodic features, management speak and the use of if to direct staff) – features that are only possible to identify as significant with the help of a corpus. The textbook also claims to incorporate insights from the Cambridge Learner Corpus in order to identify and clarify typical learner errors and to avoid teaching structures that are unproblematic for students – another example of the way that corpus data is now commonly informing language teaching.

Specialized corpora are increasingly being exploited to describe and teach academic or professional discourse, as they are better able to address the language needs of specific groups (e.g. Flowerdew Reference Flowerdew, Connor and Upton2004; Hyland Reference Hyland2008; Feak & Swales Reference Feak, Swales and Harwood2010). This trend towards smaller, specialized corpora, developed locally to meet local needs, is likely to continue, given the wide availability of electronic text, increasingly powerful computers and improved corpus software (Rundell Reference Rundell1996).

2.2 Discourse analysis

Discourse analysis (in its applied linguistics sense) has always had close ties with foreign language pedagogy, and a wide range of books exist that explore the links between theory and practice (e.g. Cook Reference Cook1989; McCarthy Reference McCarthy1991; Hatch Reference Hatch1992; McCarthy & Carter Reference McCarthy and Carter1994; Thornbury & Slade Reference Thornbury and Slade2006; Jones Reference Jones2012; Flowerdew Reference Flowerdew2013). These cover important themes (such as language variation, transactional/interactional distinctions, implicature and the cooperative principle, speech acts, text cohesion/coherence, discourse intonation and discourse structure), which can heighten language teachers’ overall discourse awareness and have positive knock-on effects in the classroom. The impact of these insights on globally marketed textbooks has been relatively modest to date, although an increasing emphasis on authentic language models (e.g. Gilmore Reference Gilmore2007a) means that relevant features are more likely to crop up in texts and can be highlighted by teachers, even if they are not the focus of attention in the original materials. Some textbooks have been more successful than others in adopting a ‘language as discourse perspective’. The Touchstone series (McCarthy, McCarten & Sandiford Reference McCarthy, McCarten and Sandiford2004, Reference McCarthy, McCarten and Sandiford2005, Reference McCarthy, McCarten and Sandiford2006) and the Natural English series (Gairns & Redman Reference Gairns and Redman2006, Reference Gairns and Redman2010), for example, cover topics such as opening, extending and closing conversation, vague language, discourse markers, listener responses, oral narratives and discourse intonation. Other books worthy of mention because of their more explicit discourse focus are Exploring spoken English (Carter & McCarthy Reference Carter and McCarthy1997), Discourse analysis in the language classroom (Riggenbach Reference Riggenbach1999), and Intonation in context (Bradford Reference Bradford1988). Exploring spoken English contains recorded samples of authentic conversational data from eight common speech genres, accompanied by detailed line-by-line commentaries which identify pervasive lexicogrammatical features often ignored in language textbooks (e.g. back-channels, deixis, ellipsis, heads and tails, vague language). Although originally intended for use, either in the classroom or for self-study, by teachers, teacher-trainers, materials developers, students of applied linguistics or advanced learners of English, it is probably most valuable as a resource to be consulted selectively by language experts, to illustrate the kinds of discourse features occurring in natural speech, and indeed there have been innumerable requests to the authors from around the world to print single chapters in teaching materials to this end (Carter, personal communication).

Discourse analysis in the language classroom describes a series of sample activities, encouraging learners to take on the role of the discourse analyst through a six-step process, which involves collecting, transcribing and analysing their own data. The teacher's role is therefore more of a facilitator, creating the conditions for learners to discover systematic features of the target language for themselves, and to ‘notice the gap’ between their own second language (L2) performance and that of more proficient speakers, in the hope that it will enhance the acquisition process (e.g. Schmidt & Frota Reference Schmidt, Frota and Day1986; Basturkmen Reference Basturkmen2001). This kind of open-ended, inductive approach, with its unpredictable conclusions, may be uncomfortable for teachers more familiar with working from a fixed syllabus, and is certainly very time-consuming (Riggenbach estimates one to five hours in class, per activity, plus a considerable amount of out-of-class work gathering and analysing data). However, the active role it gives learners in their own learning is likely to prove motivating and has been shown by some researchers to be both practical and effective (e.g. Crandall & Basturkmen Reference Crandall and Basturkmen2004). Teachers can maintain greater control over the process by providing students with materials rich in target discourse features, rather than asking them to go out and find their own, and film scenes can be a useful source of data in this kind of approach (Gilmore Reference Gilmore, Rilling and Dantas-Whitney2009a, Reference Gilmore, Mishan and Chambers2010).

Intonation in context is one of the few attempts (but see also Brazil Reference Brazil1994a and, more recently, Hewings Reference Hewings2007 or Cauldwell Reference Cauldwell2013) to translate David Brazil's systematic description of discourse intonation (e.g. Brazil Reference Brazil1985a, Reference Brazil and van Dijkb; Brazil et al. Reference Brazil, Coulthard and Johns1980) into practical pedagogic materials. Rather than associating intonation with either grammatical forms or attitude, as is commonly the case in language textbooks, it emphasizes ‘the importance of speaker choice and adjustment to the constantly changing state of play between participants in the talk’ (McCarthy Reference McCarthy1991: 114). Speakers select from options within the three sub-systems of prominence, tone and key in response to the unfolding interaction: ‘prominence’ relates to the use of vowel lengthening, increased volume or pitch variation to highlight salient words in the conversation; ‘tone’ refers to pitch changes on the tonic syllable of a tone unit which have either a ‘proclaiming’ or ‘referring’ function, depending on whether information is viewed as new or shared; and ‘key’ describes pitch jumps (between low, middle and high options) to reiterate or contrast information, or to signal topic initiation and termination in talk. This all serves to illustrate that, despite a reputation ‘for difficulty and for slipperiness’ (Brazil Reference Brazil1994b: 2), intonation can be made accessible to students in a way that can benefit both their listening and speaking proficiency, although it is rarely given the space it deserves in textbooks.

2.3 Genre analysis

Genre analysis and rhetorical consciousness-raising has had a significant impact on language learning materials. General English textbooks regularly include pre-reading/listening activities, designed to activate learners’ content or formal schemata and facilitate top-down processing and text comprehension (e.g. Carrell Reference Carrell1985, Reference Carrell1987; Richards Reference Richards1990), as well as writing exercises highlighting the structural characteristics of some of the more predictable genres, such as application letters or expository essays. Genre-based approaches have been particularly influential in English for academic purposes (EAP)/ESP contexts (e.g. Henry & Roseberry Reference Henry and Roseberry1998; Hyland Reference Hyland2002; Belcher Reference Belcher2004), where students are initiated into the rhetorical practices of their target discourse communities. Study skills books (e.g. Wallace Reference Wallace2004) attempt to develop efficient reading strategies by illustrating the systematic organization of academic textbooks or journal articles, and helping learners identify where key information can be found. Common ‘frames’ for writing assignments, such as cause/effect, problem/solution and compare/contrast, are also regularly taught to help students conform to rhetorical norms that can differ considerably from those in their first language (L1). Swales’s (Reference Swales1990) work on the structure of research articles (RAs) has been influential, and many academic writing courses cover topics such as the IMRD (introduction, method, results and discussion) model and rhetorical moves in article introductions (e.g. Swales & Feak Reference Swales and Feak1994; Hamp-Lyons & Heasley Reference Hamp-Lyons and Heasley2006). As Hyland (Reference Hyland2009: 18) points out, academic publications now play a crucial role in the careers of many researchers, so a focus on this key genre would seem justified in EAP courses:

‘Occluded genres’, which operate behind the scenes in academic communication (such as request/reminder letters to peers/supervisors or submission/review letters), have also been found to cause considerable difficulties to non-native speaker (NNS) graduate students (Swales Reference Swales1990: 77), and are beginning to receive more attention in instructional materials (e.g. Swales & Feak Reference Swales and Feak2000). Similarly, the problems posed by spoken academic genres (such as lectures, seminars, conference presentations, supervision meetings and dissertation defences) are attracting more interest from both researchers (e.g. Flowerdew Reference Flowerdew1994; Hyland Reference Hyland2009) and materials writers (e.g. Anderson, Maclean & Lynch Reference Anderson, Maclean and Lynch2004; Lynch Reference Lynch2004), as universities seek to internationalize and switch to English as their medium of instruction (Lillis & Curry Reference Lillis and Curry2010). However, Young's (Reference Young and Flowerdew1994) phasal analysis of the macro-structure of university lectures suggests that the generic descriptions (of the beginning-middle-end variety) currently provided in study skills books are over-simplistic, and there is clearly more work to be done in describing these genres and translating the findings into useful pedagogic materials.

3.1 An overview

Appendix 1 (See ‘supplementary materials, to be found at http://journals.cambridge.org/action/displayJournal?jid=lta) summarizes the findings from 44 different sources in the research literature, which explicitly address the issue of mismatches between the language models provided in textbooks and ‘authentic’ discourse found outside the classroom.Footnote ⁶ This list is, no doubt, far from comprehensive and may be biased towards my own particular research interests, but it does allow us to make some tentative observations:

1. 1. For a wide range of discourse features (including lexicogrammatical items, speech acts, generic structure, and interactional features of contingent talk), ELT textbooks often provide learners with distorted or partial representations of the target language to work from, and these are likely to impact negatively on students’ developing communicative competence.

2. 2. The work spans a time period of over three decades (1981–2012), mirroring the growing interest in discourse studies within the research community from the early 1980s onwards, and is predominantly informed by insights from DA, CA and corpus analysis. This is, perhaps, unsurprising, given that these sub-disciplines tend to focus more on the textual features of discourse and are therefore more closely aligned to the lexicogrammatical concerns of traditional language courses.

3. 3. The majority of criticisms made in the literature relate to inaccuracies in spoken, rather than written, models in ELT textbooks, and this probably reflects the difficulties materials designers have in, firstly, accessing and recording authentic speech, and then identifying pedagogically useful samples which still remain comprehensible and interesting for learners once decontextualised in the classroom (see Brown & Yule Reference Brown and Yule1983: 82; Carter & McCarthy Reference Carter and McCarthy1997: 7). Scripted dialogues, tailored to the specific goals of the materials writer and then recorded by professional actors in a studio, are an understandable choice in this situation, but native speaker intuitions about language and speech behaviour are notoriously unreliable (e.g. Slade Reference Slade1986; Wolfson Reference Wolfson1986; Burns Reference Burns1998) and, as the appendix demonstrates, the danger is that students are then presented with distorted models of the target language to learn from and mimic. A good compromise (as seen in Handford et al. Reference Handford, Lisboa, Koester and Pitt2011) is to base textbook models on authentic interactions, modifying as necessary to meet pedagogic aims, while preserving as many of the original discourse features as possible.

4. 4. The information detailed in these studies, whilst interesting, fails to provide us with any kind of systematic evaluation of the accuracy of textbook models. The CA-focused work tends to deal mainly with opening and closing sequences in conversation, while the DA- and corpus-based research covers a random selection of lexicogrammatical features; even for those target forms focused on in the research literature, there is little evidence of any subsequent impact on the design of language-learning materials. Undoubtedly, many gaps still remain in our knowledge, and it is also unclear which discourse features are more likely to impact on learners’ overall communicative competence and should therefore be prioritized in a language syllabus. Systemic functional linguistics (SFL) probably offers a better foundation for syllabus design from a discourse perspective because, rather than organizing itself around functional categories or grammatical structures, it begins by considering the typical communicative contexts and language choices available to speakers of a target group across the register variables of field, tenor and mode. Commonly occurring genres which learners are likely to encounter in their future (imagined) contexts of use can be used to structure the syllabus – any discourse features regularly cropping up in these generic samples will then have guaranteed pedagogic relevance. This is the kind of approach taken by Burns, Joyce & Gollin (Reference Burns, Joyce and Gollin1996), who use Eggins’ (Reference Eggins1990) ‘typology of spoken interactions’ to teach spoken English in the classroom. Genres illustrating differences across the tenor dimensions of (a) social distance, (b) attitude/emotion towards addressee, and (c) authority/power are presented to students and provide an effective way of systematically modelling language variation in the target discourse community (Burns Reference Burns1998).

5. 5. Given the time span covered in the survey, there is a possibility that some of the criticisms levelled at language textbooks in Appendix 1 may no longer be justified, as materials writers react to debates in the literature. Gilmore (Reference Gilmore2004), for example, identifies some improvements in the representation of spoken discourse features such as latching, hesitation devices and back-channelling in course books published between 1981 and 2001, and this might be indicative of a research-driven trend towards greater authenticity. ELT publishers are, however, notoriously conservative and slow to respond to calls for change from the applied linguistics community (Littlejohn Reference Littlejohn1992; Tomlinson Reference Tomlinson2011; Burton Reference Burton2012), and findings from research often ‘linger in the journals’ (Bouton Reference Bouton and Bouton1996) in a way that indicates a significant disconnect between the experienced realities of the two groups. As Littlejohn (Reference Littlejohn1992: 276, 277) rather cynically points out, ‘The publishers’ premises for publication [. . .] emphasise financial goals and they thus appear to have few vested interests in the precise nature of the materials (apart from cultivating continued purchasing of teaching materials)’.

6. 6. Most of the research summarized in Appendix 1 focuses on English, although there is some evidence that models for other languages exhibit similar distortions. This is unsurprising given that the method of production, using contrived language models, is likely to be the same for all foreign language textbooks.

3.2 Corpus analysis

As we saw in section 2, corpus analysis has had a significant impact on some aspects of foreign language learning, but there are at least two areas where the take-up has been less enthusiastic. Firstly, the evidence suggests that textbook authors are not yet habitually checking their materials against relevant corpus data to ensure that the language models they provide are as naturalistic and pedagogically useful as possible. As Rundell noted in 1996 (para. 28) about the ELT profession:

Very little seems to have changed in the intervening years. Burton (Reference Burton2012) asked 13 professional textbook authors about their writing practices and found that only eight of them had made any use of corpus data in the production of their materials (unfortunately, the extent of their use is not reported). The decision to refer to corpora tended to come from the authors themselves, rather than their publishers, suggesting that many publishing companies are still not actively encouraging the development of corpus-informed materials. Those participants in the survey who had never referred to corpora cited a lack of expertise, access or time as their major reasons, all of which seem rather weak from an applied linguist's perspective: the rudimentary principles of word/cluster frequency, collocation or key words are easy to grasp, a wide range of corpora are freely accessible online (see O’Keeffe & Farr Reference O’Keeffe and Farr2003: 417, 418 for an extensive list), and publishers should surely be making time to get their materials right. Overall, the picture at present is of an industry with little commitment to empirically grounded language learning materials, and it will probably require intense lobbying from teachers and learners to bring about any real change in the status quo (McCarthy Reference McCarthy2008).

A second area of limited penetration relates to the direct use of corpora by language teachers and learners themselves, and a number of researchers have commented on the widening gap between corpus-linguistics research and classroom teaching (Mukherjee & Rohrbach Reference Mukherjee, Rohrbach, Ketteman and Marko2006; Römer Reference Römer2006; McCarthy Reference McCarthy2008; Zhang Reference Zhang2008; Aijmer Reference Aijmer2009). Of course, researchers are often quick to claim practical relevance for their work and, historically, this has sometimes resulted in the over-exuberant application of (now discredited) methods such as audiolingualism in the classroom (Howatt Reference Howatt1984; Richards Reference Richards2006). However, the potential benefits for classroom-based corpus analysis, or ‘data-driven learning’ (DDL) (e.g. Johns Reference Johns, Johns and King1991), are well known: (1) DDL encourages students to discover language patterns for themselves, inductively, which is thought to lead to greater cognitive processing and deeper learning (e.g. Gollin Reference Gollin1998); (2) it is more learner-centred, allowing students to test their own hypotheses and discover rules in ways that best suit their own particular stage of interlanguage development (Rundell Reference Rundell1996); and (3) it allows students to notice patterns which have been omitted from text/reference books, either because they are too complicated or have been overlooked (Johns Reference Johns, Johns and King1991).

A number of possible causes for the disappointing levels of corpus take-up in the classroom have been put forward:

1. 1. Groom (Reference Groom2009) and Boulton (Reference Boulton, Boulton, Carter-Thomas and Rowley-Jolivet2012) suggest that it may be because of a lack of technical resources or availability of free, stable corpus analysis tools. Schools and universities are often unable to provide ready access to online corpora, or the means to display search results to the whole class, and hands-on practice by students is normally only possible in computer rooms or at home, none of which encourages spontaneous querying of corpus data. These limitations can be overcome to some extent by providing worksheet activities and printed sample concordance lines for students to analyse, which has the added advantage of removing unnecessary ‘noise’ from data (concordance lines that are difficult to re-contextualize, culturally opaque, or illustrate a different use of the target word/phrase). This is the kind of approach offered in Exploring academic English: Workbook for student essay writing (Thurston & Candlin Reference Thurston and Candlin1997), which investigates key language around important rhetorical functions in academic writing, such as referring to the literature or hedging claims.

2. 2. Another potential discourager is the time-consuming nature of inductive learning; teachers may find it impossible to make the necessary space in an already crowded syllabus and students may see corpus tasks as ‘tiresome and time-wasting activities to be endured until the teacher finally weighs in with the correct answer’ (Groom Reference Groom2009: 27). Concordance lines can be difficult to interpret at first, as the natural tendency is to read from left to right along the sentence, rather than looking for collocational patterns either side of the node (O’Keeffe & Farr Reference O’Keeffe and Farr2003). They can be particularly overwhelming for lower proficiency students, who might choose to ignore them altogether as a coping strategy (Gilmore Reference Gilmore2009b; Parise Reference Parise2009).

3. 3. On a related theme, Römer (Reference Römer2006: 128) points out that students might be reluctant to engage with corpus data because it ‘destroys their orderly world of clear-cut grammatical rules and clear right-or-wrong decisions’ – a response which is likely to be closely associated with L2 proficiency level, with more advanced learners showing increasing interest in, and willingness to engage with, the finer nuances of language variation and usage. This is in line with the concepts of scaffolding and ‘working within the Zone of Proximal Development’ (ZPD) from social constructivism, which suggest that learners need to be constantly challenged by tasks and knowledge just beyond their current level of development in order to maximize learning processes (e.g. Williams & Burden Reference Williams and Burden1997). Students are well aware of their own stage of language development and have an intuitive sense of what materials or tasks are most suitable at any particular point in time – by observing closely as learners engage in corpus-based tasks, teachers should be able to judge for themselves whether corpus tools are effectively mediating learning or not and respond appropriately.

4. 4. Technical challenges, including the mastery of unfamiliar terminology (parsing, tagging, type-token ratios, etc.) or corpus software, might also act as a deterrent (Zhang Reference Zhang2008). However, many of the online platforms available today, such as the BYU site (http://corpus.byu.edu/corpora.asp), MICASE (http://micase.elicorpora.info/) Compleat Lexical Tutor (http://www.lextutor.ca/), or the Sketch Engine (http://www.sketchengine.co.uk/), are all fairly user-friendly and well supported with online help pages or video tutorials.

5. 5. A final reason seldom mentioned in the literature is the unpredictability that a DDL approach often engenders in the classroom. Since neither the language selected by students for corpus investigation nor the resulting random sample of concordance lines pulled up to test their hypotheses can be specified in detail beforehand, teachers are forced to think on their feet, and, deprived of the sense of security which comes with having an answer key to hand, might be concerned about losing face in front of their students. The solution to this could be, quite simply, to encourage teachers and learners to see the learning process as a voyage of discovery they embark on together, on an equal footing – to accept that language descriptions presented to them in textbooks or reference books are only ever partial, that both native and non-native speaker intuitions about language can often prove to be flawed when checked against empirical data, and that the exploration of corpora can provide a more nuanced and complete picture of language patterns and meanings (Recski Reference Recski2006).

3.3 Conversation analysis

The application of insights from CA to language learning is still very much in its infancy, with interest only noticeable in the last 15 years or so (Seedhouse Reference Seedhouse2005; Barraja-Rohan Reference Barraja-Rohan2011). Although traces of it can be seen in many internationally marketed textbooks, the association is not usually explicitly acknowledged and any metalanguage from the discipline is generally avoided. Common themes cropping up include the matching of first and second pair-parts in adjacency pairs such as greetings and return-greetings (e.g. Hutchby & Wooffitt Reference Hutchby and Wooffitt2008), the use of Reactive Tokens (Clancy et al. Reference Clancy, Thompson, Suzuki and Tao1996) to show interest in listenership roles during conversation, conversation openings or closings and topic expansion to develop turns.Footnote ⁷ The only truly CA-based textbook produced to date, to my knowledge, is Beyond talk (Barraja-Rohan & Pritchard Reference Barraja-Rohan and Pritchard1997), which explores key mechanisms of spoken interaction through filmed extracts of unscripted native speaker role-plays or authentic conversation, supported by a comprehensive teacher's book and detailed transcriptions showing relevant prosodic and paralinguistic features. There were very positive reactions from students, teachers and CA researchers associated with the project (Barraja-Rohan personal communication), and although now out of print, Beyond talk is still available online (http://eslandcateaching.wordpress.com/beyond-talk/).

A number of teacher resource books are also CA-informed to some extent: Conversation (Nolasco & Arthur Reference Nolasco and Arthur1987) and Conversation and dialogues in action (Dörnyei & Thurrell Reference Dörnyei and Thurrell1992), unfortunately both out of print now, include awareness-raising activities on features such as back-channelling, projecting completion of turn-constructional units (TCUs), conversational repair and preferred/dispreferred responses to various speech acts. Also regrettable is the fact that no authentic interactions or detailed transcripts are provided in these resource books, so teachers have to generate their own materials to supplement the activities, something requiring considerable time and know-how.

Emanuel Schegloff, one of the founders of CA, has pointed out that conversation analysts are not usually overtly concerned with identifying instructable aspects of their work, and that there is ‘open terrain for enquiry [. . .] for those who will undertake to bring together the necessary training in CA with engagement with the issues which applied linguistics brings to the fore’ (Schegloff et al. Reference Schegloff, Koshik, Jacoby and Olsher2002: 18). The most recent attempt to bridge this gap between research and practice is Conversation analysis and second language pedagogy (Wong & Waring Reference Wong and Waring2010), which provides a systematic account of turn-taking and topic management in talk, sequencing practices in common speech acts (such as inviting, complementing or complaining) or oral narratives, conversation opening/closing strategies and conversational repair practices. An interesting insight to emerge from the discussion is the tendency of many EFL/English as a second language (ESL) textbooks to oversimplify language decisions in terms of register choices on formality scales, such as those shown at Figure 2, illustrating other-initiated repairs and request forms.

CA prefers to see repair mechanisms in terms of ‘their power to shine the spotlight on the trouble source’ (Wong & Waring Reference Wong and Waring2010: 229), with, for example, ‘What?’ providing a weak focus whereas an utterance such as ‘You mean + understanding check’ would give a strong focus. The authors suggest that learners should use the strongest repair mechanism possible, to clarify the source of the misunderstanding and better manage negotiation of meaning in conversation. Request speech acts, on the other hand, are viewed in terms of ‘the contingencies surrounding the granting of a request as well as their [the speaker’s] entitlement to make the request’ (Curl & Drew Reference Curl and Drew2008: 147). While modal forms, such as can or could, treat requests as non-contingent and unproblematic (with the conditions necessary for the granting of the request as already fulfilled), I wonder-prefaced forms view requests as contingent (dependent on factors that the speaker may be unsure about or which are unknown). Although the book is successful in identifying conversational features with potential pedagogic applications, the suggested classroom activities are less convincing as they tend to rely on fragments of decontextualized talk, represented with inconsistent and often unnecessarily detailed transcriptions, extracted from the research literature. No global comprehension of the texts, or the contexts which shaped them, is required before learners are asked to analyse target features, and this is something that would, in any case, be difficult to establish from such brief samples of conversation.

Figure 2

Representations of other-initiated repairs and request forms respectively

Note: adapted from Speaking naturally and New interchange – see Wong & Waring 2010: 235, 289.

The language teaching community has often reacted diffidently to ideas from CA. The reasons for this can only be speculated at, but probably stem from a general lack of awareness of the value of CA insights for developing interactional competence, a greater emphasis on written, rather than spoken, modes of communication in many parts of the world (Carter & McCarthy Reference Carter and McCarthy2006: 9), and perhaps lingering views of conversation as being unsystematic and therefore unteachable (Eggins & Slade Reference Eggins and Slade1997; Riggenbach Reference Riggenbach1999). There is still much debate as to whether direct or indirect methods of teaching conversation are more effective, particularly for ‘higher order conversational skills implicated in the global management of talk’ (Thornbury & Slade Reference Thornbury and Slade2006: 276). It seems to me, however, unlikely that the benefits of ‘noticing’ on acquisition (Schmidt Reference Schmidt1990, Reference Schmidt and Robinson2001) are limited purely to lexicogrammatical features of the language, and a number of recent studies provide support for a CA pedagogical approach in this regard (Huth & Taleghani-Nikazm Reference Huth and Taleghani-Nikazm2006; Barraja-Rohan Reference Barraja-Rohan2011; Gilmore Reference Gilmore2011).