SLA theory

SLA theory has devoted a great deal of attention to the role of cognition, dealing with issues such as whether or not there is a separate language acquisition device (LAD) or the possibility of an interface between explicit and implicit knowledge (R. Ellis 2009). Figure 23.1 is an attempt to represent the spectrum of views on the role of cognition in language learning. These range from positions that rely on an innate language acquisition device (LAD) to usage-based views, or what N. Ellis (2001: 37) refers to as a constructivist approach, which denies innate linguistic universals and claims that “the complexity is in the language, not the learning process”.

The nativist model was the basis of SLA up until the end of the 1980s (Macaro 2003) and was behind Krashen’s (1982) notions of comprehensible input and the monitor. Krashen proposed a strict distinction between uninstructed “acquisition” and instructed “learning”, leading to implicit and explicit knowledge respectively. This is often referred to as the noninterface position “that learned knowledge can never become acquired knowledge” (Doughty 2003: 258). This reduces the teacher’s role to supplying comprehensible input as instruction would not be not expected to have any effect on “acquisition” (Doughty 2003; Housen and Pierrard 2005).

Consideration of broader aspects of language learning, such as the role of perception and memory, has led many SLA theorists to conclude that there may in fact be some sort of interface allowing explicit knowledge to become implicit (Doughty 2003) and by implication that SLA is open to instruction (Housen and Pierrard 2005). There is a range of views as to how strong this interface is. R. Ellis (2006, 2009) takes a weak interface position, suggesting that explicit knowledge can in some way trigger the acquisition of implicit knowledge. There are, however, difficulties in defining and measuring explicit and implicit knowledge and learning (Ellis 2009). For example, explicit knowledge could refer to knowing the metalanguage, but it could also, and more usefully, refer to being consciously aware of how a structural feature works. Thus, different types of explicit knowledge may have different implications for the learning process. A distinction is also often drawn between declarative knowledge, knowing “that”, and procedural knowledge, knowing “how” (e.g., Macaro 2003). These views also imply some role for explicit instruction.

Those who take a usage-based approach to language deny the role of the LAD and suggest that general human cognitive learning faculties account for the learning of language, i.e., it can be viewed as learning a skill, as proposed by Anderson (1993) and DeKeyser (1998).

Those in SLA who allow for the role of explicit instruction see the importance of attention, noticing, and awareness in learning. Schmidt (2001: 3–4) suggests that “the concept of attention is necessary in order to understand virtually every aspect of SLA”. He goes on to hypothesise that “SLA is largely driven by what learners pay attention to and notice in target language input and what they understand the significance of noticed input to be”. Attention is often seen as the mechanism that controls awareness, suggesting that the level of attention may lead to different levels of awareness: perception, detection, noticing, or understanding (Housen and Pierrard 2005). Schmidt defines noticing as attending to surface features, or instances of language, as opposed to metalinguistic awareness of abstract rules or principles. Noticing involves learners in paying attention to the gap between their production and the target. Thus it is an important step in the process of acquisition. This clearly has useful implications for the classroom.

SLA has also focused on the potential role of corrective feedback (CF). While a noninterface position places no value on CF, Schmidt’s noticing hypothesis clearly allows for it and Long’s (1996) interaction hypothesis suggests that the negotiation of meaning and recasts can be a useful source of CF. Swain’s (1995) output hypothesis claims a role for CF in modified output leading to language learning.

While explicit instruction has been found to be effective, SLA theory does not provide distinctions between different types of explicit instruction (Norris and Ortega 2001) and nor does it help to identify the variables that might make one teaching technique more effective than another (Ellis 2002: 50). Housen and Pierrard (2005: 11) also note that some features are more open to instruction than others and also that “Metalinguistic rules and pedagogical descriptions can differ in clarity, intelligibility and processability so that a given target feature can be explained in both simple and elaborated terms.” Widdowson (2003: 111) discusses the differences between linguistic and pedagogic descriptions and notes that “different descriptions focus on different aspects of the truth”. This raises the question of how learners perceive both the evidence they are presented with and accompanying explanations, and leads us on to a consideration of social theories of learning.

Social theory and educational psychology

While the cognitivist views described above have mainstream acceptance within SLA, calls to acknowledge the social nature of language learning have led to a social turn in SLA (Block 2003; Ortega 2011). This movement towards a recognition of the embedded nature of learning allowing for the inclusion of a sociocultural perspective (Zuengler and Miller 2006) is often related to a participation metaphor as opposed to the acquisition metaphor used to describe much mainstream SLA (Ortega 2011). Indeed, there have been many calls for a greater focus on the social aspects of language learning (e.g., Atkinson 2011; Block 2003; Lantolf 1996; Lantolf and Thorne 2006) as there has been increased recognition that knowledge is formed through interaction with a social context (Sanz 2005).

A number of theories within the field of educational psychology are of particular interest to language teachers, e.g., Socio-Cultural Theory (SCT), Social Interactionism, and Constructivism. SCT takes into account the context of communication and views it as “emergent and cumulative based on shared knowledge with an interlocutor, and that communication involves the sending and receiving-constructing of ‘assumptions’ (rather than stand-alone ‘messages’)” (Thorne 2000: 228). Thus it views “language use in real-world situations as fundamental, not ancillary, to learning” (Zuengler and Miller 2006: 37). Lantolf and Thorne (2006: 4), in describing the role of SCT in language, argue that “because SCT is a theory of mediated mental development, it is most compatible with theories of language that focus on communication, cognition and meaning rather than on formalist positions that privilege structure”. Here “meaning” refers to conceptual meanings that mediate thinking rather than referential meaning. Vygotsky’s SCT describes how learners gain control and independence when they “appropriate mediational means, such as language, made available as they interact in socioculturally meaningful activities” (Zuengler and Miller 2006: 39). This suggests both teachers and peers have a role to play in the language development of learners.

Williams and Burden (1997) review influences on language teaching from different approaches to educational psychology. They note that constructivists, unlike behaviourists, understand that “the sense that learners themselves seek to make of their worlds, and the cognitive or mental processes that they bring to the task of learning” (1997: 12) are essential parts of the learning process. Therefore the teacher and student can co-construct meaning as they bring their subjective realities together, mediating learning and the formation of new concepts. Macaro (2003) notes the relevance of this approach in language teaching and Blyth (1997: 51) suggests its usefulness when dealing with poor textbook explanations, which present rules as if they were a “direct reflection of an objective reality”.

A further theoretical approach to come from educational psychology is social interactionism which maintains “we learn a language through using the language to interact meaningfully with other people” (Williams and Burden 1997: 39). Williams and Burden propose social interactionism, encompassing both cognitive and humanistic perspectives as “a much-needed theoretical underpinning to a communicative approach to language teaching” (1997: 39). This framework emphasizes the dynamic interaction between teacher, learner, task, and context. In the second language learning context the teacher and the learner must interact to establish meaning through effective cross-cultural communication.

Other theories that have focussed on the socially and situationally embedded nature of language learning include Atkinson’s sociocognitive approach (2011), which focuses on the integrated role of the mind, body, and world in SLA. This holistic approach is also represented in Acton’s use of haptics based on movement and touch (2013).

The idea that we co-construct discourse in the classroom underlies Swain’s (2000: 112) extension of her output model “to include its operation as a socially-constructed cognitive tool. As a tool, dialogue serves second language learning by mediating its own construction, and the construction of knowledge about itself”. More recently Swain (2006) has introduced the term “languaging” to describe this. Gibbons (2006) also analyses the bridging role of talk between teachers and students as co-constructed discourse. The theory of Intercultural Language Teaching (Lo Bianco, Liddicoat, and Crozet 1999) is also based on the importance of cross-cultural communication occurring in a third place where conceptual and experiential learning and conceptual learning can take place with the help of dialogue in developing “a shared, pragmatic understanding of what we’re talking about” (Carr 1999: 105). This supports the role of explicit instruction “using a new metalanguage which enables both teachers and learners to talk about language and culture” (Crozet and Liddicoat 1999: 121).

Phonology and L2 speech research

Having reviewed some of the theories related to language learning we will now look at theories that focus specifically on phonology and the learning and teaching of pronunciation.

Phonology involves abstract categories such as sets of segments (phonemes), tones, intonation, and voice quality (Shockey 2003). The fact that these categories are abstract is generally acknowledged, e.g., Fromkin et al. (1996). However, generative and usage-based views of phonology are in disagreement as to the nature of those categories.

Generative phonology views the categories of phonology as being determined by the presence or absence of certain distinctive features (Chomsky and Halle 1968). That is, it assumes underlying phonological rules that can be acquired through access to a Universal Grammar (UG). This implies that the teacher should focus on the physical production of sounds, i.e., the motor skills, because:

• these rules are innate and cannot be taught (Krashen 1982) and
• speech is “no more than the transmission phase of language” (Cruttenden 2001: 296), i.e., it is seen as the physical representation of language but is somehow separable from the underlying meaning.

However, as Chomsky and Halle (1968: 3) note, pronunciation is not just a matter of phonological rules, there are “many other factors as well – factors such as memory restrictions, inattention, distraction, nonlinguistic knowledge and beliefs, and so on”. Clearly the impact of “other factors” on performance is much greater for the L2 learner, which leads us to look for different theoretical positions with greater explanatory power in terms of language learning and teaching.

Bybee (2001: 34) proposes a usage-based model that “goes beyond structuralist models to show how language use gives rise to structure”. This model views pronunciation as an integral part of the meaning-making process rather than the transfer of a set of underlying phonological rules as it observes how speakers categorize language and how they relate the physical sounds to meaning. Bybee concludes that phonological categories are based on exemplars and the development of prototypes. The value of a usage-based approach is that it focuses on meaning and not on rules that create the impression of dichotomous features such as the voiced/voiceless distinction, when in fact they overlap (Mompean 2014). The implications for the teacher and learner are that it is possible through cognitive skill learning processes to help learners understand the relationship between sound and meaning. The difficulties in understanding this relationship are explored in the first tip for teachers, where we consider the implications of the gap between phonology and the physical sounds of speech.

L2 speech research helps in understanding speech perception and provides us with a number of insights into category formation processes. Kuhl and Iverson’s (1995) perceptual magnet effect suggests that sounds are perceived in terms of the prototype categories for the language or languages we know. We then assimilate nonprototypical members into the prototype and shrink the acoustic-phonetic space towards it. Therefore “L1 prototypes constrain learners’ abilities to perceive contrasts in L2 by the ‘pulls’ they exert” (Leather 1999: 5). Flege (1990: 255), in describing what is required for the process of speech learning to be successful, notes that one must be able to “establish central perceptual representations for a range of physically different phones (‘sounds’) which signal differences in meaning, and [develop] motoric routines for outputting sounds in speech production”. As well as a prototype for each sound category, there is evidence for episodic effects, that is, one remembers the particular examples one has heard and these are called upon when categorizing sounds (Pisoni and Lively 1995).

Flege’s (1995) Speech Learning Model (SLM) provides useful insights into L2 pronunciation learning. It “aims to account for age-related limits on the ability to produce vowels and consonants in a native-like fashion” (1995: 237). It is assumed that “our phonetic systems remain adaptive over the life span and reorganise to allow for L2 sounds by adding new phonetic categories or modifying old ones” (1995: 233). To do this we must discern some of the phonetic differences and be able to relate the L1–L2 sounds at an allophonic level. With age we may find it harder to notice subtle but possibly significant differences and classify similar L1–L2 sounds as being the same. This model allows for the influence of the L1 and for the impact of an L2 on modifying existing categories and forming new ones. Therefore it directs our attention to the processes of category formation and the role that training may play.

While the SLM focuses on the formation of categories, another approach is to focus on how the stream of sound can be interpreted by the listener to recognize words and the phonemes that make them up psychologically. Lively, Pisoni, and Goldinger (1994: 265) explain that this is complex for a number of reasons. Firstly, there is a lack of acoustic–phonetic invariance, i.e., acoustic forms of words and phonemes are different when produced by different speakers, but they are also different when produced by the same speaker in different occurrences, in different situations, and phonetic contexts. Secondly, phonemes are not produced linearly; they overlap and are co-articulated, making it impossible to reliably map acoustic features to perceived phonemes. Finally, there is a lack of segmentation, which means we have to rely on context-sensitive cues such as stress and intonation contours to assist.

One of the major difficulties in learning a new category is to discern those aspects of the auditory signal that mark the sound as belonging to that particular category. This may lead to learners attending to a difference that is of little significance to the native speaker. A number of researchers have noted how difficult it can be to get learners to focus on the “right” cues (e.g., Munro and Bohn 2007). However, Guion and Pederson (2007) investigated the role of attention in phonetic learning and concluded that explicit directing of attention can help adult learners to discern novel phonetic contrasts better.

Cognitive linguistics and phonology: a pronunciation learning and teaching framework

In reviewing theories that have the greatest explanatory power from the perspective of what goes on in the classroom, I have found Cognitive Phonology (CP), derived from Cognitive Linguistics, provides a coherent explanation for the phenomenon of adult L2 pronunciation learning. CP’s conclusions, although not always arrived at in the same manner, are congruent with those of many of the theories reviewed above: pronunciation is a cognitive skill that can be learned using our general learning faculties as proposed by the interface position from SLA theory; pronunciation learning is situationally embedded involving a complex interplay of social and cognitive variables in the construction of meaning as suggested by SCT and other socially oriented theories that adopt a participant metaphor. Lantolf (2011) notes that SCT fits in comfortably with Cognitive Linguistics; CP also takes a usage-based approach to phonology and is in line with many of the findings from L2 speech research.

Cognitive Phonology is a branch of Cognitive Grammar, situated within Cognitive Linguistics, stemming largely from the work of Langacker (1987) and Taylor (2002). Cognitive Linguistics is based on the premise that the cognitive abilities required for language are similar to those required for other learning and “it argues that language is embodied and situated in the sense that it is embedded in the experiences and environments of its users” (Mompean 2006: vii). It uses what is known about cognition to build theories of language acquisition rather than the other way around and it totally rejects the Chomskyan view in Generative Theory that language is in the mind and autonomous (Taylor 2002). It also rejects computer processing type analogies for the way the brain processes language and is distinct from cognitive psychology, which focuses more on subconscious processing on inaccessible mental representations (Anderson 2000).

Mompean (2014) reviews the main implications of Cognitive Linguistics for the understanding of phonology in terms of two guiding assumptions. The first is “language, including phonology, is the outcome of properties of cognition” (2014: 357). He analyses the importance of three cognitive abilities with relevance for phonology: categorization, perception, and conceptual combination. Categories are considered central to conceptual and linguistic organization but most notably Cognitive Linguistics does not accept the traditional view of categories as being discrete and defined by necessary and sufficient conditions with features distributed evenly across category members. Rather, categories are defined through “overlapping similarities with different category members or similarity to a central or prototype member of the category” (2014: 360). Perception is also a cognitive ability. It enables us to recognize similarity, which leads to categorization. Another relevant perceptual capacity is attention to salience, which enables us to distinguish between the figure, or what needs to be heard as prominent, and ground, which does not require our attention. The implications of this for teachers and learners are taken up in the next section.

The second assumption of Cognitive Linguistics is “linguistic organisation (phonological inclusive) is also the outcome of the bodies humans have and how they interact with the sociophysical world” (Mompean 2014: 357). This focus on the social and cultural aspects of language and language learning is in line with the social turn seen within SLA.

Teaching tip one: understand all is not as it seems

Applying theory from cognitive phonology to the classroom situation, the first thing we need to remember is that our phonological concepts determine how we categorize sounds and that these concepts are language-specific. That is, we perceive speech differently so when learning L2 pronunciation we have to learn a different way of thinking about sounds. As teachers we have to remember that when we think about English pronunciation we are thinking about it through a filter built up through many years, maybe even a lifetime, of experience in extracting meaning by categorizing sound into relevant categories. We are so proficient at it that it is easy to forget that the actual sounds produced do not relate one-to-one with the phonemes we see in the dictionary. So there is a two-step process: first we have to understand the difference between what we actually say (the physical sounds) and what we think we say (the phonology, or the way we categorize sounds). Then we have to help students to go through the same process with their first language by getting them to use their ears to move away from the way they are used to thinking about sounds so that we can help them to understand the way English speakers think about sounds.

As Shockey (2003: 10) notes, “most people speaking their native language do not notice either the sounds that they produce or the sounds that they hear”. Shockey also reports that often whole sounds are omitted even though the listener still perceives them. Phonologically it is easy to think of pronunciation as a sequence of one sound after the other. However, acoustically it has been demonstrated that this is not in fact the case. Warren (1982) provides experimental evidence to show that speech perception is not dependent on an ability to identify component sounds and their orders, that in fact a great deal of speech would be too fast to do this, and that we rely on holistic pattern recognition.

Fraser (2004), taking the example of the words bat and bad, provides a good demonstration of the difference between what we think we hear and the physical sounds and the implications of this for teachers. While most naïve listeners will say the difference between bat and bad is that one has a /t/ and the other has a /d/, in fact the greatest difference is in the length of the vowel. Acoustically the /d/ and /t/ are surprisingly similar. Of course this understanding of phonemes is also often supported by spelling, which misleads both teachers and learners through what is known as a literacy bias (Linell 2005). This is one simple example of the difference between what we perceive and the actual sounds, and one can see the significance of this in a teaching context. If a teacher insists the learners produce a /d/ or a /t/ we are likely to see the unexpected production of aspiration or an additional vowel (referred to as epenthesis) at the ends of words.

An analogy of how speech perception works can be seen through visual perception puzzles. Take, for example, the picture in Figure 23.2. When you look at the picture, you first need to understand that there is another way of looking at it. You might see a young lady. I can explain this by saying I see an old lady. However, this might not help you to see it, it just tells you to keep looking. I can try and explain what to look for, to see the old woman. She has a big nose. She’s looking down. Her mouth is a thick black line. If this doesn’t help, I could put the salient lines into the foreground and push the other ones into the background by drawing the outline of the old woman’s nose and eye (as in Figure 23.3). I may also have to think of other ways to make the second perspective clearer to you. Now you should be able to see what the differences are. Of course when you look again later, you may still have difficulty in finding the second perspective.

This is an example of how figure ground organization works, which demonstrates the sorts of difficulties our students might be having when trying to adjust their figure ground perception. As was noted in the review of L2 speech research, it can be difficult to get students to notice the salient cues. Cognitive phonology provides the explanation that as in different languages different aspects of incoming acoustic data are phonologically salient, one has to learn what is salient in the target language in order to form the concepts required for the L2’s phonological categories. As teachers, it is important to understand this if we are going to be effective in helping learners to form new concepts.

Therefore the first tip is to make sure you can move away from your subjective perception of L1 speech. It does not feel subjective because that is how your speech community perceives it, which in turn makes it feel like reality, or the truth. Therefore we need to untruth, or step outside, our normal view of the world. If we do not do this and instead hold to our perceptions as being the objective truth of what those sounds are, we are in danger of insisting to our students that they need to produce and hear certain sounds that are not in fact there. If we can recognize the physical sounds and the gap between them and the perceived sounds, we are in a better position to help our students make the connection between form and meaning. While the more teachers understand about phonology the better, the main point is to learn to be able to step aside from your own perceptions, or to untruth, and listen to sounds more neutrally to try and imagine how they might sound to your students.

Teaching tip two: generate dialogue

Given the different phonological perceptions of teachers and students, effective communication about pronunciation requires that we establish common understandings with our students. For example, when talking to students about syllables it is easy to think of it as a category and forget that the concept behind this category varies from language to language. This leads to the situation where we fail to communicate with our students because on the one hand English conceives of a syllable as containing at least one vowel, which may have a number of consonants in the onset and coda. On the other hand, other languages have different concepts of syllables, such as consisting of one consonant followed by one vowel. We may not be consciously aware of these concepts but they control how we perceive speech. These cross-linguistic differences in concepts may mean that if my L1 follows a consonant vowel syllable structure I am likely to relate the syllable to the presence of a consonant, so that every time I come across a consonant I will expect a vowel to follow. They may also render talking to students about syllables useless (Couper 2006) unless we find ways to communicate about them. This involves the development of a common way of talking, or the social construction of metalanguage. I observed the value of co-constructing such a language through a number of classroom interactions and found empirical evidence for its effectiveness in improving pronunciation (Couper 2011). I refer to this language as socially constructed metalanguage (SCM).

Socially constructed metalanguage (SCM) refers to the kind of metalanguage that is needed for effective metalinguistic communication. Such communication, as with all cross-cultural communication, relies on both parties having a common understanding of the concepts that are being discussed. SCM requires the teacher and the learners to work together to construct common ways of talking about these concepts. This involves the teacher in understanding how the learners interpret the sounds of the target language. One way the teacher can do this is by asking learners to describe the difference between two productions. Equally it involves the learners in understanding how the sounds they produce are interpreted by the native speaker. It is social in the sense that it is owned by the class as a group and it refers to the social nature of language learning and the role of social construction of meaning. Once this metalanguage has been developed, it can be used throughout the course for quick and effective feedback. While the term SCM has been developed in relation to teaching pronunciation, it could just as easily apply to the use of explicit instruction in all aspects of language teaching.

In practice, this means that we need to start by asking our students to tell us how they understand the sounds of English. For example, if a student says fishy when they want to say fish, I write the two words on the board. I will explain that, to my ears, it sounded like fishy, pointing to and underlining the difference. I can then model the two words, asking the student to tell me how they are different. They are unlikely to say there is an extra syllable; rather, they will suggest the “shy” in fishy is longer or louder. Alternatively, they might suggest the “sh” in fish is shorter or quieter. This tells me that while I perceive an extra syllable, they simply perceive it as a different way of saying the same sound. This means we need to help the learners understand the salient differences between the two for the proficient English speaker, in other words to establish the phonological category.

To do this, I will ask the student to say both words and I will point to the one I hear. In giving them feedback, I can use the language they have already used to describe the differences between the sounds. So I might tell them to make the “sh” shorter or quieter to help them produce fish rather than fishy. Of course they still need a great deal of practice, especially if they have been saying it incorrectly for a long time. However, once the learner understands how these two sounds are categorized differently by English speakers they can remind themselves what they have to do to get the message across.

By engaging in this sort of dialogue with our students we can focus on all aspects of pronunciation. For example, a recent study (Couper 2012) focused on word stress, beginning with learners’ current perceptions of word stress in both their own languages and English and helping them to understand how the concept of stress was actually different in different languages. Again, by having this classroom conversation, common understandings were developed and communication was more effective when providing explanations and feedback. What it amounts to is effective cross-cultural communication, enabling teachers and learners to better understand each other and develop a common basis on which to build language proficiency.

While this approach was developed directly from cognitive phonology, theories that focus on language awareness and social theories such as SCT might also support this approach. What it offers over the traditional SLA approach is that it defines a specific type of explicit instruction that it suggests is better than any old type of explanation. Indeed, Couper (2011) provides convincing empirical evidence to support this claim. With regards to the need for explicit as opposed to implicit instruction, some of us may learn many of these L2 phonological concepts implicitly, but there will usually be some concepts that require explicit intervention. This is where the teacher needs to be aware of the need to provide this sort of instruction.

Teaching tip three: establish category boundaries through critical listening practice

As we learn L2 phonological concepts we are learning how sounds are categorized in the L2. As a part of this process we need to establish the prototypical sound as well as all the variations that would be classified as belonging to the same category. Fraser (2009) suggests critical listening as a way of helping learners in this process. Critical listening involves the learner in listening for the contrast between two productions: one that is acceptable and one that is not. Typically there should be a meaningful difference, and ideally it would involve comparing the learner’s production when it is acceptable with when it is not. As with SCM, it involves helping learners to understand how the sounds are perceived by the native speaker. It involves a focus on developing speech perception and learning where the boundaries are between the different phonological categories. Again, this approach is derived from cognitive phonology.

In practice this might involve learners recording themselves and then listening to their recording and comparing it with a model in conjunction with getting feedback from peers or the teacher. Even though this has the potential to be face threatening when done with a large class, I have found that as long as you can develop the appropriate supportive atmosphere the students will work cooperatively to help establish these category boundaries. This helps not just the particular learner but all other learners as well as they can learn more examples of what belongs to these categories. It is important to note that these exercises should focus on meaningful differences rather than what might be construed as slight differences in accent. The question is whether the target language speaker interprets the sounds as intended.

Fraser (2009: 301) provides evidence to support the claim that critical listening, focusing “on the contrast between a correct (or appropriate) pronunciation versus an incorrect (or inappropriate) pronunciation within a particular communicative act”, could help in forming new phonological concepts, in this case establishing the boundaries between the /r/ and /l/ categories in English. Further support was found by Couper (2011) in teaching learners to produce syllable codas without epenthesis, or producing an additional vowel.

This technique is also supported by findings from L2 speech research, which has clearly shown that adults can be trained through comparing and contrasting to learn these categories and their boundaries. Strange (1995: 40) reviews research into the effect of training and concludes adults can learn new phonological contrasts as they “retain the auditory perceptual abilities that are required for the detection and discrimination of the acoustic parameters that carry phonetically relevant information”, i.e., the right kind of training can help adult learners to improve their L2 speech perception. Rochet (1995) also concludes that difficulties adults have in perceiving L1–L2 differences in similar phonemes is not representative of a sensory-based loss but rather of a change in selective attention.

One way to help set up critical listening practice is to get students to record themselves at the beginning of a course and use this as a diagnostic giving them initial feedback and encouraging them to set up their own goals for improving their pronunciation. The teacher can use these recordings to prepare examples for critical listening that contrast different productions of the same word or utterance.

Teaching tip four: meaningfully integrate pronunciation into further practice activities

Cognitive linguistics assumes that form is motivated by meaning and the way we relate meaning to form is determined by our perception (Holme 2012). We have already seen how SCM and critical listening are important in the formation of L2 phonological concepts. However, to fully form and automatize these concepts we often need further practice. While SCM and critical listening involve meaningful and genuine communication about language and the relationship between form and meaning, we also need to help learners use this language in communicative situations. Here, teachers can draw on their experience with communicative language teaching to devise activities that will support the development of phonological concepts.

For example, having observed my learners’ difficulties with syllable codas such that drunk and drunker sounded the same to them I developed an information gap activity, which would help to make the difference in meaning and form clear and give them the opportunity to practise and receive feedback multiple times. I called this the “Drunk Snail Game” and it involves sets of cards containing pairs of adjectives and their comparatives, describing animals such as: a drunk snail/a drunker snail; a loud parrot/a louder parrot, etc. I printed each item on a separate card with an appropriate picture taken from clip art. The object of the game is to find matching pairs by correctly pronouncing what is on the card. Another player who has the matching card then has to correctly pronounce what is on that card. The players check that they have understood each other by comparing their cards. If they are the same, they have succeeded and, if not, one of them will realize that they pronounced it incorrectly. The details of this game are described in Couper (2014). This is an example of how we can develop activities that focus on concept formation by establishing appropriate figure-ground organization, helping learners to establish category boundaries through the cognitive ability to compare and detect discrepancy and learn from feedback, and providing multiple experiences in a context that presents learning as social behavior.

Most other communicative activities can also be structured with the aim of helping learners to form and practise new concepts. For example, in setting up a role-play activity we might first consider the type of language that will be needed and possibly have some controlled practice with the language beforehand, that is, the lexico-grammatical aspects as well as the phonological ones. One could record the role-play and review it with the class, allowing them to focus on their performance and discussing certain pronunciation features that caused misunderstandings. This might then lead on to further practice once learners have understood the form-meaning relationship. Another common task for learners is to give oral presentations. Again, we can prepare for this by focusing on certain key features such as phrasing and sentence stress. Then after their performance, one could review the video leading on to increased awareness as well as plans for further practice. These sorts of activities are already common in the communicative language classroom. We can easily integrate a pronunciation focus by thinking about what helps learners to practise and form new concepts.

Teaching tip five: provide corrective feedback focused on concept formation

Corrective feedback (CF) is the most common way in which teachers engage with pronunciation (Foote et al. 2013); therefore it is important to consider how it is provided. A key factor in determining the effectiveness of CF is the extent to which the learner understands the correction. The first step then is to make sure learners understand that it is a correction and that they understand precisely where the problem lies. As an example of how easy it is for corrective feedback to miss the mark, Couper (2013: 10) reports the following event:

In the practice stage, during which key words from the listening were being practised, Ay repeated “exports” for “experts”. The teacher then repeated “experts” several times while Ay continued to say “exports”. This was being done without any visual support on the board. Bea then explained to Ay that the /p/ changes to /b/ and Ay commented that “when I say my name everyone thinks it’s a ‘p’ but it’s a ‘b’”. The teacher finally realized that Ay had been focusing on the wrong thing (i.e. the pronunciation of /p/ rather than the second vowel in “experts”) and wrote the two words on the board, underlining the vowel. Ay then saw where the problem was and attempted to correct it, although she still found it difficult.

This reminds us of the need to write things on the board to help learners see where their mistakes are. Of course, even when they know where the problem is it does not always mean they can fix it. Cognitive linguistics makes it clear that if learners do not understand the phonological concept they are not likely to learn from the correction. So we have to ensure that we communicate effectively with our students when making corrections. This is where SCM comes in. If we have already developed this in relation to the particular phonological concept that is the focus of correction, then we should be able to quickly communicate what the problem is. For example, when correcting syllable codas, rather than referring to the additional syllable, or even the additional vowel, we can use the learners’ descriptions such as “that’s too strong”, “say it shorter”, “it becomes quiet”, “make it smaller” (Couper 2013: 9). If they are still unable to correct it then they need further practice, possibly using critical listening techniques, to fully establish the concept. In conclusion, the teacher must focus on providing CF that contributes to the formation of phonological concepts.

Lyster, Saito, and Sato (2013) in their review of research into oral CF note that there is general support for CF from a range of theoretical perspectives; however meta-analyses of research into the effectiveness of CF tend to categorize it according to whether it is explicit or implicit, a prompt or a reformulation, or whether it is form focused or meaning focused. The position put forward in this paper is that these variables fail to isolate the most important variable, which is the degree to which the communication between the teacher and student is actually understood.