Volume 66 Issue 6, December/décembre 2010, pp. 907-933

Abstract

In recent years, researchers have become increasingly interested in the notion of spontaneous focus on form (FonF) (i.e., attention to linguistic forms that arise incidentally during meaningful communication). Previous research has indicated that such FonF is beneficial for L2 learning. However, this research has focused mainly on reactive FonF, with relatively little investigation into pre-emptive FonF. This paper presents findings from a large-scale, classroom-based study that examined the role of both reactive and pre-emptive FonF in second-language development as measured by tailor-made, student-specific post-tests. Data from 54 hours of classroom interaction within seven intact ESL classes at three levels of language proficiency revealed that both reactive and pre-emptive FonF occurred. However, pre-emptive FonF led to higher individualized post-test scores than did reactive FonF. Furthermore, the amount, type, and effectiveness of FonF were strongly related to the learners’ level of language proficiency.

Résumé

Au cours des dernières années, l'idée d'une centration spontanée sur la forme (à savoir une attention secondaire aux formes de la langue dans le contexte d'une communication signifiante) a fait l'objet d'un intérêt croissant chez les chercheurs. Une recherche antérieure avaient montré qu'une telle centration sur la forme favorise l'apprentissage de la L2. Cette recherche, cependant, portait essentiellement sur la centration réactive sur la forme et très peu sur la centration préventive. Les résultats présentés ici ont été obtenus lors d'une étude d'envergure effectuée en classe qui portait sur l'effet des deux sortes de centration sure la forme dans le développement de la langue seconde, mesuré au moyen de post-tests conçus expressément pour des étudiants et pour l'expérience. Les données tirées de 54 heures d'interaction en classe avec sept groupes complets d’étudiants d'ALS de trois niveaux de compétence différents révèlent que les deux types de centration, réactive et préventive, se sont manifestés. Toutefois, la centration préventive a donné lieu à de meilleurs résultats individuels au post-test que l'attention réactive. En outre, le degré, le type et l'efficacité de l'attention accordée à sur la forme étaient étroitement corrélés au niveau de compétence linguistique des apprenants.

Dissatisfaction with traditional form-focused instruction, on the one hand, and the inadequacies of purely meaning-focused approaches, on the other, have led many second language acquisition (SLA) researchers to argue that attention to linguistic forms must occur within the context of meaningful communication. This is what has come to be known as focus on form (FonF) (Doughty, 2001; Doughty & Williams, 1998; Ellis, 2001; Fotos & Nassaji, 2007; Long, 1991, 2000; Long & Robinson, 1998; Nassaji & Fotos, 2004, 2007). The notion of FonF was introduced by Long (1991), who distinguished it from a focus on forms (FonFs) and a focus on meaning. FonFs comprises traditional approaches that involve planned selection and presentation of linguistic rules and structures in isolated manners (Long & Robinson, 1998). It is based on a synthetic syllabus and the assumption that language consists of a series of grammatical forms that can be acquired sequentially and additively. Focus on meaning is an analytic approach (Wilkins, 1976) that characterizes learning in terms of mere exposure to communicative language use with no effort to draw learners’ attention to form. It is based on the assumption that learners are able to process and analyze the language to which they are exposed and arrive at its underlying grammar inductively. FonF is also concerned with meaning; however, it differs from a focus on meaning in that it also attempts to draw learners’ attention to linguistic forms.

No research has directly examined the difference between a FonF and a FonFs.1 However, it has been suggested that FonF is more effective than either a FonFs or a focus on meaning. Long (2000) argues that FonF is particularly effective because it occurs at the time when learners need it, is tailored to interlanguage development, and is attuned to the learning process. The significance of FonF also derives from the belief that learners learn linguistic items more effectively when they attend to them while their primary attention is on meaning (Ellis, Basturkmen, & Loewen, 2001a, 2001b; Nassaji & Fotos, 2004). Because L2 learners have difficulty in attending to both form and meaning at the same time due to limited processing capabilities, they are considered to learn from opportunities that raise their attention to form while they are processing input for message content (VanPatten, 1990, 1996). Doughty and Varela (1998) suggested that FonF techniques are potentially effective because their aim is ‘to add attention to form to a primarily communicative task rather than to depart from an already communicative goal in order to discuss a linguistic feature’ (p. 114). In this sense, the proposed benefit of FonF derives from the cognitive support provided by the overall context of meaning and also from the fact that learners notice the linguistic form at the time when they need it for communication (Doughty & Williams, 1998).

Two types of noticing have been distinguished in the field of SLA, both of which have been hypothesized to assist language acquisition: noticing the gap and noticing the hole. Noticing the gap occurs when learners notice a difference between their current interlanguage output and the target-like production of the forms needed for communication (e.g., Doughty, 2001; Gass, 1997; Williams, 2005). Noticing the hole occurs when learners produce output and then realize that they are not able to say what they want to say in the target language (Swain, 1993, 1995, 2005). FonF is suggested to promote opportunities for both noticing the gap and noticing the hole. For example, when learners participate in interaction and then receive interactional FonF, such as recasts, on their erroneous utterances, they may compare their original output with the correct form in the feedback and realize that their production differs from the target-like production. Thus, they may notice a gap in their linguistic knowledge. This then triggers cognitive processes involved in learning, leading to the restructuring of the learners’ interlanguage (Doughty, 2001). Similarly, when the teacher or an interlocutor requests clarification from learners during communication, learners may be pushed to make their output more accurate or appropriate. By being pushed to output, learners may recognize that they are not able to say what they want to say. Thus, they notice a hole in their interlanguage, an observation that may promote a subsequent search for solutions and consequently modified output, which is considered essential for language development (Swain, 1995; Swain & Lapkin, 1995).

In his discussion of form-focused instruction, Ellis (2001) made a distinction between two types of FonF: incidental and planned. Incidental FonF consists of instruction that involves no pre-selection of the target form. In this kind of FonF, attention to form occurs spontaneously in the course of meaningful communication. Planned FonF entails treatment of pre-selected target forms. Thus, in this sense, planned FonF looks like FonFs, but the difference is that in the former, treatment occurs while the learner's primary focus is on meaning whereas in the latter the focus is exclusively on form. Examples of planned FonF include communicative tasks that contain many exemplars of a target form (i.e., input flood), textual enhancement that involves attempts to highlight instances of certain target forms in written or oral texts, or form-focused tasks that are designed to elicit from learners particular linguistic forms required for completing the task.

According to Ellis (2001), FonF can occur both reactively and pre-emptively. Reactive FonF refers to attention to form that occurs when the teacher or an interlocutor responds to a learner's erroneous utterance during meaningful interaction. Thus, it provides opportunities for negative evidence. Pre-emptive FonF occurs when the learner's attention is raised to a particular form anticipated to be problematic irrespective of any actual error. Such attention to form can be initiated by either the teacher or the learner (Ellis et al., 2001a). An example of a teacher-initiated FonF is when students are performing a communicative task and the teacher believes that a certain linguistic form is problematic at a particular moment of the discussion. In such cases, the teacher may decide to address the problem pre-emptively by taking time out from the communicative task to discus or explain that form. Similarly, when doing the same activity, the students may initiate a FonF by posing a question about a form they do not know or have difficulty with. According to Ellis et al. (2001a), both reactive and pre-emptive FonF are problem oriented. However, they are different in the nature of the problem. Whereas reactive FonF is prompted by an actual problem in the learner's oral or written production, pre-emptive FonF involves potential problems.

Given the theoretical importance attributed to FonF, an increasing number of studies have recently examined its usefulness in communicative contexts. However, most of these have investigated planned FonF in experimental or quasi-experimental research (e.g., Ammar & Spada, 2006; Doughty & Varela, 1998; Han, 2002; Leeman, 2003; Lyster, 2004; Lyster & Izquierdo, 2009; Mackey, 1999; Mackey & Philp, 1998; Sheen, 2007). A few studies have also investigated incidental FonF (Ellis et al., 2001a, 2001b; Loewen, 2005; Lyster & Mori, 2006; Lyster & Ranta, 1997; Nassaji, 2007, 2009; Williams, 1999, 2001), most of which focused on reactive FonF in response to learner production errors. To date, relatively few studies have examined the occurrence and effectiveness of pre-emptive FonF (Ellis et al., 2001b). Therefore, we know much less about how pre-emptive FonF takes place in L2 classrooms and how effective it is than we do about other types of FonF. This is surprising given that pre-emptive FonF is also a kind of incidental FonF that occurs spontaneously within a broader communicative context, hence providing opportunities for integrating form and meaning in instructional contexts.

One study that examined incidental FonF in the L2 classroom was conducted by Williams (1999) and investigated the occurrence of learner-generated attention to form among eight adult ESL learners doing collaborative group work. Learner-generated attention to form was identified through language-related episodes (LREs), defined as occasions during which learners talked or asked about the language they or other learners used during pair or small-group work. Williams found that learner-generated attention to form existed in the classroom context but was not very frequent, particularly at lower levels of language proficiency. Using the same database, Williams (2001) examined the effectiveness of student-generated attention to form by means of tailor-made, learner-specific tests and found a strong relationship between such attention to form and learners’ performance on subsequent post-tests.

Ellis and his colleagues have examined incidental FonF in a series of descriptive studies carried out in adult ESL classrooms in a private English school in New Zealand (e.g., Basturkmen, Loewen, & Ellis, 2004; Ellis et al., 2001a, 2001b; Loewen, 2004, 2005; Loewen & Philp, 2006). Ellis et al. (2001b) examined the use of incidental FonF in two upper-intermediate adult ESL classrooms. Analyzing 12 hours of meaning-focused instruction, the researchers found that incidental FonF occurred quite frequently in the form of both reactive and pre-emptive FonF. They also examined the frequency of learner uptake and found that it occurred in response to pre-emptive FonF and was also quite frequent (71%). But it occurred more frequently in student-initiated than in teacher-initiated FonF (84% vs. 27%). Ellis et al. (2001a) examined learner uptake in both pre-emptive and reactive FonF, including recast and negotiation strategies. The researchers found that uptake occurred very frequently in response to both types of FonF, but successful uptake occurred most frequently in response to reactive FonF, particularly after recasts (76%). This rate was much higher than that found in previous research of incidental feedback in classroom contexts, such as in Lyster and Ranta (1997), which examined the occurrence of corrective feedback in French immersion classrooms and found that uptake occurred in response to only 41% of recasts, with only 18% being successful. Ellis et al. attributed this difference to the context of research, as Lyster and Ranta's study took place in young French immersion classes whereas their research took place in an adult intensive ESL program.

Loewen (2005) collected and analyzed data from 17 hours of meaning-focused lessons to examine the relationship between incidental FonF and language learning. To measure learning, Loewen used individualized tests based on episodes in which the learner had participated and given to the same learner one day after the interaction and then two weeks later. Loewen found that learners were able to perform accurately on the post-test items at least 50% of the time. Loewen also found a significant relationship between successful uptake following FonF and learners’ test scores. In a separate analysis of the data, Loewen and Philp (2006) examined the occurrence and usefulness of reactive FonF including recasts, elicitation, and metalinguistic feedback. They found that learners were able to recall the correct form targeted by feedback in subsequent post-tests between 53% (for recasts) and 75% (for elicitations). Their results also showed that the effectiveness of recasts depended on their formal characteristics and that these characteristics differentially affected both learners’ uptake and the accuracy of their post-test scores. For example, whereas characteristics such as stress and declarative intonation, one change, and multiple feedback moves were predictive of successful uptake, interrogative intonation, shortened length, and one change were predictive of the accuracy of learners’ post-test scores.

In a more recent study, Zhao and Bitchener (2007) investigated the occurrence of incidental FonF in learner–learner and teacher–learner interaction in about 10 hours of classroom interaction in two intact classes in an English language program in New Zealand. The study found that in both patterns of interactions FonF occurred frequently. It also found significantly more reactive FFEs in teacher–learner interactions than in learner–learner interactions. The study also compared learner uptake following FonF in the two types of interactions. No difference was found between the two interactional contexts (53.3% vs. 52.8%), suggesting that learner–learner and teacher–learner interactions were equally effective in terms of learner uptake. This study is significant because it indicates that incidental FonF also occurs in learner–learner interactions. However, since the study used uptake to measure the effectiveness of FonF, it did not address the extent to which incidental FonF affected L2 learning.

The above studies provide important insight into how incidental FonF works in L2 classrooms. However, as can be seen, most of these studies, particularly those of pre-emptive FonF, were conducted in the same instructional context (i.e., private English language schools in New Zealand). Thus, more studies are needed to investigate the usefulness of such FonF in other contexts. Studies of FonF have shown that the effectiveness of FonF depends on the context of instruction. Sheen (2004), for example, found that learners benefited more from reactive FonF such as recasts in EFL contexts, where the learners were more oriented to attending to linguistic forms, than in content-based communicative contexts. Furthermore, previous studies have measured the effectiveness of FonF mainly in terms of learners’ uptake. Of course, as noted earlier, Loewen (2005) examined the effectiveness of FonF for L2 learning by using tailor-made, individualized post-tests. However, his research focused on reactive and student-initiated FonF, not teacher-initiated pre-emptive FonF. Williams (2001) also examined the effectiveness of attention to form using individualized post-tests. However, she focused mainly on reactive FonF. In particular, she excluded all instances of teacher-initiated pre-emptive FonF. As Williams explained, although she used the term initiation, by initiator she meant the individual who responded to a learner's error in a reactive FonF episode (either the teacher or the learner), not the one who triggered the error or initiated a pre-emptive FonF sequence. Williams called this ‘a rather unusual use of the term Initiator’ (p. 329).

The current study set out to investigate the occurrence and effectiveness of FonF in an adult intensive ESL program in Canada. Using a large database of naturalistic classroom interaction, the study examined and compared the usefulness of both reactive and pre-emptive FonF, including student- and teacher-initiated FonF. Reactive and pre-emptive FonF may contribute to language learning differently because of a difference in their degree of explicitness and also in the nature of learner involvement.

As noted earlier, previous studies of incidental FonF have examined student-initiated FonF but not teacher-initiated FonF. The reason might have been because teacher-initiated FonF takes place based on the teacher's prediction that the learner may have difficulty with a certain linguistic form. In such cases, it is possible that the learner may already know the targeted form. If so, the FonF is not addressing a gap in the learner's knowledge. The current study examined the effectiveness of both student- and teacher-initiated FonF. However, attempts were made to ensure that the FonF examined addressed a gap in the learner's knowledge (see the methodology section for details). Data were collected at three levels of language proficiency: beginner, intermediate, and advanced. Thus, the study also examined the differential impacts of language proficiency on the occurrence and usefulness of FonF. Language proficiency has been shown to play an important role in the effectiveness of FonF (e.g., Ammar & Spada, 2006; Iwashita, 2001; Mackey & Philp, 1998; Williams, 2001). However, except for Williams’ study on student-generated FonF, these have focused mainly on planned (reactive) FonF in experimental research rather than on incidental FonF. In trying to extend previous research in this area, the current study also examined the extent to which learners’ level of language proficiency affects the occurrence and usefulness of different types of incidental FonF.

The following research questions were examined for the purpose of the study:

  1. How often does reactive versus pre-emptive FonF occur in adult ESL classrooms?

  2. What effects does reactive versus pre-emptive FonF have on L2 learning?

  3. Is there any relationship between the amount and effectiveness of different types of FonF and learners’ levels of language proficiency?

The research is part of a large-scale project that investigated the role of spontaneous FonF in adult L2 classrooms. The study was conducted in an intensive English language program in a university context in Canada, consisting of 12 weeks of English instruction with 22 hours of instruction per week. The program offers six levels of instruction, from upper-beginner to advanced levels. The classes are held both in the morning and in the afternoon and are normally taught in two-hour lessons. The goal of the program is to foster fundamental communicative skills for beginner and intermediate level learners and to provide opportunities to develop language skills further at the more advanced levels. Learners are placed in class levels based on an in-house writing and speaking language placement test administered by the program at the beginning of each semester.

Participants

Seven intact classes taught by five different teachers at three levels of language proficiency (upper beginner, intermediate, and advanced) participated in the study. The number of students in the classes observed ranged from 12 to 18, with a total of 105 students. The age of the students ranged from 18 to 48 with a mean of 23.18 years (SD = 4.68). They were from a variety of language backgrounds, including Chinese, Japanese, Korean, Portuguese, Spanish, Russian, Taiwanese, Thai, Arabic, and Turkish. By the time of the study, the participants had been in Canada from 1 to 14.5 months. The teachers were native speakers of English, with teaching experience ranging from 8 to 22 years. They all held bachelor degrees in disciplines such as education, linguistics, applied linguistics, and English literature, and they also had additional ESL qualifications. Two had TESL certificates and three had diplomas in applied linguistics. The teachers had taught ESL at different levels.

Procedure

A total of 54 hours of communicative lessons were analyzed (18 hours for each level). The data come from 35 lessons over a period of two 12-week semesters. The classes were video- and audio-recorded using a digital recorder with a wireless clip-on microphone attached to the teacher. The lessons were observed and recorded by non-participant observers who were trained graduate research assistants. When observing the classes, they took detailed field notes about the nature of each lesson, the activities, students’ participation, and patterns of interactions. Such information was used when coding the data and developing student-specific individualized tests.

In all the classes observed, the teacher's instructional approach was mainly communicative, with an emphasis on developing the students’ listening and speaking abilities. The classes also included occasional grammar and vocabulary exercises. However, the analysis concentrated mainly on meaning-focused activities, that is, on those activities that hold meaningful communication rather than teaching pre-selected isolated forms (Loewen, 2005) as a primary goal. These activities took various forms and included classroom tasks conducted individually, in pairs, or by groups of students, such as one-way or two-way communicative tasks, general class discussions or debates, activities in which students discussed issues and topics encountered in their reading, listening, or other classroom tasks, or those nominated by the teacher.

Data analysis

To analyze the data, the recorded classroom data were first fully transcribed using normal orthography. To identity instances of FonF, the data were first searched for the presence of what was called ‘FonF episodes’ (FFE). Following Ellis et al. (2001a), an FFE was defined as a piece of discourse in which the learners’ attention was drawn briefly to form by either the teacher or the learner during a communicative activity. The onset of an FFE was marked by the discourse element that triggered the FonF, such as a learner's production error in reactive FonF or the teacher's or a learner's query or statement about a linguistic form in pre-emptive FonF. The FFE ended with a change in topic or another FonF. The following is an example of an FFE taken from the data in the study:

S1 The Japanese mafia?
S2 Yeah, they are not allowed to get into the public bath.
S3 Really?
S2 Yeah, of course they can't.
T Not allowed to go into them.
S2 Yeah, go into.

Trained research assistants carefully reviewed the recordings and written transcriptions and identified all the FFEs for analysis. Each analyzed a different portion of the data but finally a sample of 90 minutes of the data was coded by two of them independently and there was a 90% agreement between the two raters. After identifying the FFEs, the assistants coded each episode in detail in terms of types of FonF and their various characteristics. Initially two broad categories of FonF were identified: reactive and pre-emptive.

Reactive FonF was defined as an occasion on which the teacher responded to a learner's problematic utterance during meaning-focused interaction using various forms of interactional strategies such as recasts, clarification requests, repetitions, elicitations, or explicit correction. The following is an example of a reactive FonF:

S Sometimes they complain that.
T They complain ABOUT that.
S Yeah, they complain about that.

Pre-emptive FonF involved instances in which the teacher or learner initiated attention to form by posing a query or providing comments about a linguistic form. Following Ellis et al. (2001a), two types of pre-emptive FonF were distinguished: teacher initiated and student initiated. A teacher-initiated FonF occurred when the teacher drew the learner's attention to form by either asking a question or providing information about a form:

T Do you know what ballet is?
SS Yes.
T What's ballet?
S1 Ah, the classical music, the ladies …
T Men and women, it's dance, it's a type of dance. Ballet.

Student-initiated FonF, on the other hand, occurred when a student initiated the FFE by raising a question or requesting information regarding a linguistic form:

S Do you say driving licence or driver's licence?
T Driver's licence.
S Because on my driver licence it says, it says driving licence.
T Yeah, but it's from Sweden, right?
S [laughs].

The reliability of the coding was established by comparing a sample of 20% of the FFEs jointly coded by two independent raters. Kappa coefficients obtained for general categories of FonF (reactive and pre-emptive) was 0.95 and for different types of pre-emptive FonF (teacher-initiated and student-initiated) was 0.97.

Testing

Due to the unplanned nature of incidental FonF, designing pre-test/post-test studies is not feasible for investigating such FonF in naturalistic classroom contexts. Therefore, following previous research, to measure the effectiveness of FonF tailor-made, student-specific post-tests were used. These tests were developed based on FFEs in which individual students had participated and were then given to the same student who had been involved in the FFE one week after each FFE. An involved student was the one who (1) had made a production error in a reactive FFE, (2) had initiated an FFE in a student-initiated pre-emptive FonF, or (3) was directly addressed by the teacher and/or responded to the teacher's query in a teacher-initiated FonF.2

As noted earlier, previous research has not examined the effectiveness of teacher-initiated FonF, possibly because such FonF may target linguistic forms that learners already know. In other words, it might not address a gap in learners’ knowledge. In the case of reactive and student-initiated FonF, a gap of knowledge can be assumed because the learner has either made an error or has raised a question regarding a particular linguistic form. However, in the case of teacher-initiated FonF, such a gap cannot be always assumed. Thus, when testing, attempts were made to ensure that the examined teacher-initiated FonF also addressed a gap in learners’ knowledge. To this end, only teacher-initiated episodes that met the following criteria were included in the analysis: a student had responded to the FonF but the response was not correct, or the student had explicitly indicated that he or she did not know the targeted form. Instances when the teacher made a statement or raised a question regarding a particular form but immediately followed that question with an answer or explanation were not included in the testing. The rare cases in which information requested in a student-initiated FonF was not supplied were also excluded from the testing. Of course, it is hard to make a clear distinction between knowing and not knowing a language form. Gaps in knowledge are also a matter of degree and can relate to the extent to which students have control over a particular form (Nassaji, 2009). For example, the learner might have declarative knowledge of a target form but be unable to use it properly in spontaneous discourse.

The tests included a variety of formats, depending on the nature of the target form, including fill in the blanks, multiple choice, error identification/correction, definition matching, and picture-cued tests. Pronunciation errors were tested in individual sessions in which the student was presented with the targeted item and asked to read and pronounce it out loud. The following is an example of an FFE and a test item that was developed based on that episode.

S1 I don't know wolf.
S2 Wolf is a … wolf is a [gesturing with hands] …
T Wolf [opening and closing hand like a mouth of a dog]. Ah … similar to a dog. A wild dog is a wolf.

To score learners’ responses to the test items, a strict coding criterion was adopted in which the responses were scored as correct only if the response was fully correct. Incorrect responses included those that involved a wrong answer or no answer, repeated the error that had occurred during the initial interaction, or introduced new errors. In cases in which the learner had to produce the correct form but the form provided was different from the items targeted by the feedback, they were initially coded as substitute-correct but were then combined with the other correct responses because they also involved correction. Since there were a small number of such responses, their addition was shown to make little difference to the final results. All of the responses were checked and scored by two independent raters and an inter-rater agreement of 97% was established.

TEST ITEM Which animal is a wolf?

For statistical analysis, since the data involved frequency data chi-square tests were used but only when the results were compared across class levels. No chi-square statistics were used when comparisons were made within the same class or in the whole database because in such cases the assumption of independence of observation required for chi-square is violated. Independence of observation is violated when the data that are used in different cells of a contingency table come from the same learner or learners (Hatch & Lazaraton, 1991). When the analysis involves data from the same class or groups of learners, it is possible that the same learner has contributed data to the different cells of the contingency table.

A total of 1,986 FFEs were indentified and coded in the 54 hours of classroom data analyzed. Therefore, the occurrence of FFEs was one every 1.63 minutes, ranging from one every 1.4 and 1.5 minutes at the beginner and advanced levels to 2.3 minutes at the intermediate level. The overall rate of FonF replicates what was found in Ellis et al.'s (2001a, 2001b) research, in which they reported 448 FFEs in 12 hours of meaning-focused lessons (one FFE every 1.6 minutes).

The first research question concerned the occurrence of different types of FonF. Table 1 displays the proportion of reactive and pre-emptive FonF, as well as of student- and teacher-initiated FonF in the whole database. As can be seen, pre-emptive FonF occurred far more frequently than reactive FonF. Of all the FFEs indentified, 1,373 (69%) were pre-emptive and 613 (31%) were reactive. Of the two types of pre-emptive FonF, 60.5% was teacher initiated and 39.5% was student initiated.

Table

TABLE 1 Occurrence of different types of FonF across the database

TABLE 1 Occurrence of different types of FonF across the database

Frequency %
Reactive 613 31.0
Pre-emptive 1,373 69.0
Total 1,986 100.0
Student initiated 543 39.5
Teacher initiated 830 60.5
Total 1,373 100.0

With regard to different types of FonF across class levels, pre-emptive FonF occurred more frequently than reactive FonF at all levels, but the proportion of reactive FonF was higher in advanced classes than in beginner classes (Table 2). The chi-square test, however, showed no significant difference among class levels, χ2 (2, N = 1,986) = 4.44, p = 0.108, suggesting that the distributions of reactive and pre-emptive FFEs were similar across levels.

Table

TABLE 2 Occurrence of reactive and pre-emptive FonF across class levels

TABLE 2 Occurrence of reactive and pre-emptive FonF across class levels

Pre-emptive Reactive Total
Frequency % Frequency %
Beginner 553 71 221 29 774
Intermediate 337 69 148 31 485
Advanced 483 66 244 34 727
Total 1,373 69 613 31 1,986

As for different types of pre-emptive FonF (Table 3), there were more instances of teacher-initiated than student-initiated FonF. However, their proportions differed across class levels, with student-initiated FonF occurring more frequently in the beginner level (43%) than in the advanced classes (34%), suggesting that beginner students were more involved in initiating a FonF episode than advanced level students. The chi-square test showed that the relationship between class level and types of pre-emptive FonF was statistically significant, χ2 (2, N = 1,373) = 8.51, p = 0.014.

Table

TABLE 3 Occurrence of student- and teacher-initiated pre-emptive FonF across class levels

TABLE 3 Occurrence of student- and teacher-initiated pre-emptive FonF across class levels

Student initiated Teacher initiated Total
Frequency % Frequency %
Beginner 237 43 316 57 553
Intermediate 140 42 197 58 337
Advanced 166 34 317 66 483
Total 543 40 830 60 1,373

The next analyses examined the effectiveness of FonF. Altogether, a total of 768 FFEs were tested, out of which 385 were pre-emptive and 383 were reactive. Table 4 shows the frequency and percentage of learners’ correct responses to these test items. As can be seen, out of the total number of FFEs tested, learners responded correctly to 444 (58%) of them in the subsequent post-tests. Therefore, learners were able to respond correctly to more than half of the linguistic items tested. This result is in line with that reported by Loewen (2005), which showed that learners responded accurately to at least 50% of the test items. As for the difference between reactive and pre-emptive FonF, overall, pre-emptive FonF led to a higher rate of correct test scores than reactive FonF (63% vs. 53%). As for student- versus teacher-initiated FonF, student-initiated FonF resulted in a much higher rate of correct test scores than teacher-initiated FonF (72% vs. 46%). These findings suggest that learners benefited more from FonF when it was pre-emptive and also when they initiated it themselves.

Table

TABLE 4 Post-test accuracy scores of types of FonF across the database

TABLE 4 Post-test accuracy scores of types of FonF across the database

Frequency Number correct % correct
Pre-emptive 385 241 63
Reactive 383 203 53
Total 768 444 58
Student initiated 247 177 72
Teacher initiated 138 64 46
Total 385 241 63

Tables 5 and 6 display the results of the post-test scores for reactive versus pre-emptive FFEs within and across class levels. As can be seen in Table 5, overall there is an increase in the learners’ correct test scores as we move from the beginner to the advanced levels (48% to 59% and then to 66%). The chi-square test showed a significant difference across class levels, χ2 (2, N = 768) = 17.23, p < 0.001, suggesting that the advanced level learners benefited significantly more from FonF than did the less advanced learners. The results also showed an interaction between class level and types of FonF. Although pre-emptive FonF resulted in higher rates of accurate test scores than reactive FonF at the beginner and intermediate levels (56% vs. 41% and 65% vs. 53%, respectively), at the advanced level, reactive and pre-emptive FonF resulted in almost the same rate of correct responses (66% and 65%). Thus, while beginner and intermediate learners benefited more from pre-emptive than reactive FonF, advanced level learners benefited equally from both reactive and pre-emptive FonF.

Table

TABLE 5 Post-test accuracy scores of reactive and pre-emptive FonF within class levels

TABLE 5 Post-test accuracy scores of reactive and pre-emptive FonF within class levels

Frequency Number correct % correct
Beginner Pre-emptive 108 60 56
Reactive 133 55 41
Total 241 115 48
Intermediate Pre-emptive 127 83 65
Reactive 131 69 53
Total 258 152 59
Advanced Pre-emptive 150 98 65
Reactive 119 79 66
Total 269 177 66
Table

TABLE 6 Post-test accuracy scores of reactive and pre-emptive FonF across class levels

TABLE 6 Post-test accuracy scores of reactive and pre-emptive FonF across class levels

Frequency Number correct % correct
Pre-emptive Beginner 108 60 56
Intermediate 127 83 65
Advanced 150 98 65
Total 385 241 63
Reactive Beginner 133 55 41
Intermediate 131 69 53
Advanced 119 79 66
Total 383 203 53

Table 6 displays the results for each type of FonF across class levels. As the table shows, the intermediate and advanced learners scored higher on both pre-emptive and reactive FonF than did beginners. However, the difference was not statistically significant for pre-emptive FonF, χ2 (2, N = 385) = 3.17, p = 0.204. But it was statistically significant for reactive FonF, χ2 (2, N = 383) = 15.80, p < 0.001. These findings suggest that learners at all levels benefited comparably from pre-emptive FonF, but they benefited differently from reactive FonF, with advanced level learners benefiting significantly more from reactive FonF than learners at lower levels.

The next analyses compared the frequency and percentages of students’ correct test scores for student-initiated versus teacher-initiated FonF within and across class levels. As Table 7 shows, in all classes, student-initiated FonF resulted in higher rates of correct test scores than teacher-initiated FonF. However, the difference is largest in the beginner classes. While at the advanced level, 56% of the teacher-initiated FFEs led to correct test scores compared to the 72% of student-initiated FFEs, at the beginner level only 28% of teacher-initiated FFEs led to correct test scores compared to the 69% of student-initiated FFEs. Furthermore, while student-initiated FFEs were fairly, and almost equally, successful at all levels, with the rate of correct scores ranging from 69% to 73% (Table 8), the classes differed significantly in term of accuracy scores for teacher-initiated FonF, χ2 (2, N = 138) = 7.25, p = 0.027. These results suggest that whereas learners benefited almost equally from student-initiated FonF at all levels, they benefited much less from teacher-initiated FonF at the beginner level than at the advanced.

Table

TABLE 7 Post-test accuracy scores of student-initiated and teacher-initiated FonF within class levels

TABLE 7 Post-test accuracy scores of student-initiated and teacher-initiated FonF within class levels

Frequency Number correct % correct
Beginner Student initiated 72 50 69
Teacher initiated 36 10 28
Total 108 60 56
Intermediate Student initiated 86 63 73
Teacher initiated 41 20 49
Total 127 83 65
Advanced Student initiated 89 64 72
Teacher initiated 61 34 56
Total 150 98 65
Table

TABLE 8 Post-test accuracy scores for student- and teacher-initiated FonF across class levels

TABLE 8 Post-test accuracy scores for student- and teacher-initiated FonF across class levels

Frequency Number correct % correct
Student initiated Beginner 72 50 69
Intermediate 86 63 73
Advanced 89 64 72
Total 247 177 72
Teacher initiated Beginner 36 10 28
Intermediate 41 20 49
Advanced 61 34 56
Total 138 64 46

In general, FonF occurred at the rate of one episode every 1.63 minutes in the 54 hours of communicative lessons examined. This rate is comparable with that found in Ellis et al. (2001a), where FFEs occurred at the rate of one every 1.6 minutes. Therefore, it confirms the observation that incidental FonF does occur in ESL communicative classes and that it occurs rather frequently. With regard to the occurrence of different types of FonF, pre-emptive FonF occurred far more frequently than reactive FonF (twice as often). This suggests that the teachers in this study tended to be more proactive than reactive in their strategies to address learners’ linguistic problems. In other words, they did not simply wait for learners to produce errors to which to respond. Rather they used other opportunities during meaning-focused activities to draw learners’ attention to form. These strategies seemed to work well and can be considered appropriate given that learner-negotiated interactions are typically limited in adult classroom settings (e.g., Foster, 1998; Pica, 2002).

As for the two types of pre-emptive FFEs, the majority were teacher initiated rather than student initiated. This might be an indication of teachers’ control of the patterns of communication in the classroom (Johnson, 1995). It might also be a reflection of their sensitivity to and awareness of learners’ linguistic needs in the course of interaction and their attempts to anticipate and resolve the learners' problems ahead of time. However, although there were fewer instances of student-initiated than teacher-initiated FonF in general, there were significant variations across class levels, with student-initiated FFEs occurring more often in the beginner classes. One reason for the lower occurrence of student-initiated FonF at the higher level might be that learners at higher levels of language proficiency do not feel as much need for assistance as do lower-level learners who, because of their more limited linguistic knowledge, may require and hence ask for more assistance. However, the lower incidence of student-initiated FonF at the advanced level contrasts with Williams's (1999) finding that advanced level learners were more likely to initiate a FonF. One possible explanation for this difference might be related to the nature of interactions observed. Williams’s findings were mainly based on data from learner–learner interactions during collaborative group work. This study, however, also involved teacher–student interactions. It is possible that the occurrence of FonF is affected by different interactional patterns. Such a difference was indeed reported by Zhao and Bitchener's (2007) study, which found a higher frequency of student-initiated FonF in learner–learner interactions than in teacher–learner interactions (40.7% vs. 5.6%). Explaining this difference, Zhao and Bitchener pointed out that learners are more willing to ask questions of each other than of their teacher.

As for the effectiveness of FonF, the results indicated that students responded correctly to more than half of the FFEs tested. This result is similar to that found by Loewen (2005), who, using similar tests, found that learners were able to respond correctly to post-test items at least 50% of the time. They are also similar to that found in Williams's (2001) work, which showed that learners performed accurately on episode-related post-test items at least 45% of the time. Taken together, these findings suggest that incidental FonF may have important beneficial effects for language learning in the classroom. Of course, since both the current study and those previous studies tested the effects of FonF using untimed individualized post-tests, learners’ accurate responses to the test items do not guarantee that learners are also able to use the forms spontaneously in communicative discourse. This is a limitation and suggests that future research is needed to address the extent to which incidental FonF can contribute to the development of communicative competence.

As for the difference between reactive and pre-emptive FonF, the analysis showed that overall, pre-emptive FonF was more successful than reactive FonF. The superiority of pre-emptive FonF highlights its usefulness in L2 classrooms and suggests that for incidental FonF to be effective it does not necessarily need to be reactive (cf. Long, 1991). The advantage of pre-emptive FonF over reactive FonF could result from a number of factors, such as the more explicit nature of such FonF. Pre-emptive FonF usually involves direct queries or metalinguistic comments or explanations. Reactive FonF, however, typically involves less direct feedback, such as recasts or negotiations in the course of interaction. Such feedback can be considered more implicit in nature, although the degree of explicitness and the effect vary depending on how the feedback is provided (Nassaji, 2007, 2009). When FonF is more direct and explicit, students are more likely to notice the linguistic forms. Furthermore, the study had more instances of pre-emptive FonF targeting lexical forms than of reactive FonF doing so. Lexical forms are by nature more explicit than grammatical forms and, therefore, are more likely to be noticed when they become the target of feedback, even when the feedback is relatively implicit, such as recasts (Mackey, Gass, & McDonough, 2000).3 This can also explain why pre-emptive FonF was found to be more effective than reactive FonF in the current study.

Of course, because pre-emptive FonF is fairly explicit, one may wonder whether such FonF contributes to the development of implicit knowledge. This question was not addressed in the study because the tests may have favoured learners’ explicit knowledge. Learners could have used their conscious attention when responding to the items.4 However, there are reasons to believe that the kind of explicit knowledge that metalinguistic information typically presents in pre-emptive FonF can contribute to the acquisition of implicit knowledge. Explicit knowledge not only promotes noticing but also helps learners to monitor their production and hence produce accurate output, which can then serve as auto-input to the implicit knowledge system (Ellis et al., 2001b). Explicit knowledge can also lead to the production of unanalyzed language that learners may incorporate into their implicit interlanguage grammar later, when they are developmentally ready (Spada & Lightbown, 2008). In addition, in a recent study, Ellis, Loewen, and Erlam (2006) found empirical evidence that explicit metalinguistic feedback contributes to the development of implicit knowledge and that its contribution is greater than implicit feedback such as recasts.

With respect to the different kinds of pre-emptive FonF, the current study found that student-initiated FonF led to a much higher rate of correct test scores than teacher-initiated FonF. The superiority of student-initiated FonF highlights the importance of the role of the learner in FonF and confirms Williams’s (2001) findings. The superiority of student-initiated FonF could be because the attention to form arises out of learner need. Therefore, learners are more motivated to learn (Ellis et al., 2001b). Also, when learners ask for assistance, they may pay more attention to the subsequent input, which would consequently help them better notice the targeted form. Furthermore, the learner's need in student-initiated FonF may also be an indication that the learner is cognitively ready to learn the targeted form (Williams, 2001).

Finally, the analysis showed a strong relationship between the effectiveness of FonF and language proficiency. It showed that as learners’ language proficiency increased, the effectiveness of incidental FonF also increased, and this was true for both reactive and pre-emptive FonF. Furthermore, the results showed an interaction between language proficiency and types of FonF. Although beginner and intermediate learners benefited more from pre-emptive FonF than reactive FonF, advanced learners benefited equally from the two types. Also, advanced learners benefited significantly more from reactive FonF than did less advanced learners. If we assume that reactive FonF is more implicit than pre-emptive FonF, the advantage of reactive FonF for advanced level learners can be taken to indicate that language proficiency is an important factor to the effectiveness of implicit feedback but not to the effectiveness of explicit feedback. One reason for this might be that explicit knowledge that results from explicit feedback is not influenced to the same degree by learners’ developmental constraints or readiness as is implicit knowledge (Ellis, 2005). Furthermore, in view of skill acquisition theories, as learners become more proficient, their automaticity in language use will also develop (e.g., DeKeyser, 1996). Due to higher levels of automaticity, advanced learners can have more attentional resources available to devote to FonF. Thus, they may be better able to notice the targeted form than less advanced learners.

However, it should be noted that while this study investigated different categories of FonF, it did not examine the exact nature of FFEs or how they were realized in the classroom. Thus, it is not clear what features of the FonF or interaction influenced their effectiveness. More qualitative analysis of the data designed to provide an in-depth account of what actually goes on during FFEs is needed. Furthermore, this study was descriptive, not experimental. Therefore, it did not include a control group. Because of this, it is not possible to definitively determine that the effects shown for different types of feedback actually result from the feedback. Of course, it should be noted that conducting experimental research on incidental FonF is difficult because of the lack of pre-selected target forms and also the fact that neither the feedback nor the way in which it is provided is planned. Other factors might have played a role in the effects of FonF, including the overall context of instruction, the type of linguistic forms focused upon, and even the learners’ age. For example, this study was conducted in an adult intensive ESL program. It is possible that pre-emptive FonF does not occur to the same degree in more content-based programs or in those that involve younger learners. Furthermore, reactive FonF is less disruptive to the flow of communication than pre-emptive FonF because it involves negotiation and hence can keep the focus on content (Williams, 2005). Thus, teachers may favour reactive over pre-emptive FonF in content-based classrooms. Also, since pre-emptive FonF usually involves metalinguistic comments, teachers may prefer not to use it with children as children do not usually possess the kind of knowledge needed to understand metalinguistic explanations. For the same reason, it is also possible that children may benefit less from pre-emptive FonF than adults do. These observations are speculative, however, and need to be empirically investigated in future research.

Overall, this study adds to those that have investigated the nature and usefulness of incidental FonF in L2 communicative classrooms. The findings confirm that incidental FonF occurs both reactively and pre-emptively in L2 communicative classrooms. In addition, the findings showed positive effects for both reactive and pre-emptive FonF on learning the targeted forms, as measured by individualized post-tests. These effects can be taken to support the claim that spontaneous attention to form in the context of meaning-focused activities facilitates L2 acquisition (e.g., Doughty, 2001; Ellis et al., 2001a; Long, 2000; Long & Robinson, 1998). The results, however, showed that of the two types of FonF, pre-emptive was more effective than reactive in improving learners’ post-test scores. This finding suggests that to be effective, FonF does not need to be in response to learner errors in the course of interaction. It can also occur irrespective of errors. Furthermore, the findings showed that learners benefited differently from different types of FonF depending on their level of language proficiency. In particular, the findings showed that language proficiency had a greater influence on the effectiveness of reactive FonF as opposed to pre-emptive FonF.

Acknowledgements

This study was supported by a research grant from the Social Sciences and Humanities Research Council of Canada (SSHRCC). I would like to thank the ESL teachers and students who participated in this study as well as all the graduate research assistants who assisted with different phases of the research.

Notes

1 Norris and Ortega's (2000) meta-analysis included a comparison of studies with different types of instructional treatments. However, none of the studies they examined had directly compared a FonF with a FonFs or focus on meaning.

2 To find out whether FonF had any effects on enhancing the learning of those who were observing the FFE, the tests were also administered to students who were not directly involved in the episodes but were simply observers. The analysis in this article, however, focuses on students who were directly involved.

3 Of course, as Ellis (2006) points out, it is difficult to establish with certainty what kind of knowledge learners use when performing a language test.

4 This explicit nature of the tests might be another reason why pre-emptive FonF led to better results.

Ammar, A., Spada, N. (2006).One size fits all? Recasts, prompts, and L2 learning.Studies in Second Language Acquisition.28,543-574doi:10.1017/S0272263106060268 Google Scholar
Basturkmen, H., Loewen, S., Ellis, R. (2004).Teachers' stated beliefs about incidental focus on form and their classroom practices.Applied Linguistics.25,243-272doi:10.1093/applin/25.2.243 Google Scholar
DeKeyser, R.M. (1996).Exploring automatization processes.TESOL Quarterly.30,349-357doi:10.2307/3588151 Google Scholar
, Doughty, C.J.Robinson, P. (2001).Cognitive underpinning of focus on form.Cognition and second language instruction.Cambridge:, Cambridge University Press206-257 Google Scholar
, Doughty, C.J., Varela, E.Doughty, C.J., Williams, J. (1998).Communicative focus on form.Focus on form in classroom second language acquisition.Cambridge:, Cambridge University Press114-138 Google Scholar
Doughty, C.J., Williams, J. (1998).Focus on form in classroom second language acquisition.Cambridge:Cambridge University Press Google Scholar
Ellis, R. (2001).Introduction: Investigating form-focused instruction.Language Learning.51,1-46doi:10.1111/j.1467-1770.2001.tb00013.x Google Scholar
Ellis, R. (2005).Principles of instructed language learning.System.33,209-224doi:10.1016/j.system.2004.12.006 Google Scholar
Ellis, R. (2006).Current issues in the teaching of grammar: An SLA perspective.TESOL Quarterly.40,83-107doi:10.2307/40264512 Google Scholar
Ellis, R., Basturkmen, H., Loewen, S. (2001a).Learner uptake in communicative ESL lessons.Language Learning.51,281-318doi:10.1111/1467-9922.00156 Google Scholar
Ellis, R., Basturkmen, H., Loewen, S. (2001b).Preemptive focus on form in the ESL classroom.TESOL Quarterly.35,407-432doi:10.2307/3588029 Google Scholar
Ellis, R., Loewen, S., Erlam, R. (2006).Implicit and explicit corrective feedback and the acquisition of L2 grammar.Studies in Second Language Acquisition.28,339-369doi:10.1017/S0272263106060141 Google Scholar
Foster, P. (1998).A classroom perspective on the negotiation of meaning.Applied Linguistics.19,1-23doi:10.1093/applin/19.1.1 Google Scholar
Fotos, S., Nassaji, H. (2007).Form focused instruction and teacher education: Studies in honour of Rod Ellis.Oxford:Oxford University Press Google Scholar
Gass, S.M. (1997).Input, interaction, and the second language learner.Mahwah, NJ:Erlbaum Google Scholar
Han, Z.H. (2002).A study of the impact of recasts on tense consistency in L2 output.TESOL Quarterly.36,543-572doi:10.2307/3588240 Google Scholar
Hatch, E., Lazaraton, A. (1991).The research manual: Design and statistics for applied linguistics..New York:Newbury House Google Scholar
Iwashita, N. (2001).The effect of learner proficiency on interactional moves and modified output in nonnative–nonnative interaction in Japanese as a foreign language.System.29,267-287doi:10.1016/S0346-251X(01)00015-X Google Scholar
Johnson, K. (1995).Understanding communication in second language classrooms..Cambridge:Cambridge University Press Google Scholar
Leeman, J. (2003).Recasts and second language development: Beyond negative evidence.Studies in Second Language Acquisition.25,37-63doi:10.1017/S0272263103000020 Google Scholar
Loewen, S. (2004).Uptake in incidental focus on form in meaning-focused ESL lessons.Language Learning.54,153-188doi:10.1111/j.1467-9922.2004.00251.x Google Scholar
Loewen, S. (2005).Incidental focus on form and second language learning.Studies in Second Language Acquisition.27,361-386doi:10.1017/S0272263105050163 Google Scholar
Loewen, S., Philp, J. (2006).Recasts in adults English L2 classrooms: Characteristics, explicitness, and effectiveness.Modern Language Journal.90,536-556doi:10.1111/j.1540-4781.2006.00465.x Google Scholar
, Long, M.H.De Bot, K., Ginsberg, R., Kramsch, C. (1991).Focus on form: A design feature in language teaching methodology.Foreign language research in cross-cultural perspective.Amsterdam:, John Benjamins39-52 Google Scholar
, Long, M.H.Lambert, R.D., Shohamy, E. (2000).Focus on form in task-based language teaching.Language policy and pedagogy: Essays in honor of A. Ronald Walton.Philadelphia:, John Benjamins179-192 Google Scholar
Long, M.H. (2006).Problems in SLA..Mahwah, NJ:Erlbaum Google Scholar
, Long, M.H., Robinson, P.Doughty, C.J., Williams, J. (1998).Focus on form: Theory, research and practice.Focus on form in classroom language acquisition.Cambridge:, Cambridge University Press15-41 Google Scholar
Lyster, R. (2004).Differential effects of prompts and recasts in form-focused instruction.Studies in Second Language Acquisition.26,399-432doi:10.1017/S0272263104263021 Google Scholar
Lyster, R., Izquierdo, J. (2009).Prompts versus recasts in dyadic interaction.Language Learning.59,453-498doi:10.1111/j.1467-9922.2009.00512.x Google Scholar
Lyster, R., Mori, H. (2006).Interactional feedback and instructional counterbalance.Studies in Second Language Acquisition.28,269-300doi:10.1017/S0272263106060128 Google Scholar
Lyster, R., Ranta, L. (1997).Corrective feedback and learner uptake: Negotiation of form in communicative classrooms.Studies in Second Language Acquisition.19,37-66doi:10.1017/S0272263197001034 Google Scholar
Mackey, A. (1999).Input, interaction and second language development: An empirical study of question formation in ESL.Studies in Second Language Acquisition.21,557-587doi:10.1017/S0272263199004027 Google Scholar
Mackey, A., Gass, S.M., McDonough, K. (2000).How do learners perceive interactional feedback?.Studies in Second Language Acquisition.22,471-497doi:10.1017/S0272263100004022 Google Scholar
Mackey, A., Philp, J. (1998).Conversational interaction and second language development: Recasts, responses, and red herrings?.Modern Language Journal.82,338-356doi:10.2307/329960 Google Scholar
Nassaji, H. (2007).Elicitation and reformulation and their relationship with learner repair in dyadic interaction.Language Learning.57,511-548doi:10.1111/j.1467-9922.2007.00427.x Google Scholar
Nassaji, H. (2009).Effects of recasts and elicitations in dyadic interaction and the role of feedback explicitness.Language Learning.59,411-452doi:10.1111/j.1467-9922.2009.00511.x Google Scholar
Nassaji, H., Fotos, S. (2004).Current developments in research on the teaching of grammar.Annual Review of Applied Linguistics.24,126-145doi:10.1017/S0267190504000066 Google Scholar
, Nassaji, H., Fotos, S.Fotos, S., Nassaji, H. (2007).Current issues in form-focused instruction.Form-focused instruction and teacher education: Studies in honour of Rod Ellis.Oxford:, Oxford University Press7-15 Google Scholar
Norris, J., Ortega, L. (2000).Effectiveness of L2 instruction: A research synthesis and quantitative meta-analysis.Language Learning.50,417-528doi:10.1111/0023-8333.00136 Google Scholar
Pica, T. (2002).Subject-matter content: How does it assist the interactional and linguistic needs of classroom language learners?.Modern Language Journal.86,1-19doi:10.1111/1540-4781.00133 Google Scholar
Sheen, Y. (2004).Corrective feedback and learner uptake in communicative classrooms across instructional settings.Language Teaching Research.8,263-300doi:10.1191/1362168804lr146oa Google Scholar
, Sheen, Y.Mackey, A. (2007).The effects of corrective feedback, language aptitude, and learner attitudes on the acquisition of English articles.Conversational interaction in second language acquisition: A series of empirical studies.Oxford:, Oxford University Press301-322 Google Scholar
Spada, N., Lightbown, P.M. (2008).Form-focused instruction: Isolated or integrated?.TESOL Quarterly.42,181-120 Google Scholar
Swain, M. (1993).The output hypothesis: Just speaking and writing aren't enough.The Canadian Modern Language Review.50,158-164 LinkGoogle Scholar
, Swain, M.Widdowson, H.G., Cook, G., Seidlhofer, B. (1995).Three functions of output in second language learning.Principle and practice in applied linguistics: Studies in Honour of H.G. Widdowson.Oxford:, Oxford University Press125-144 Google Scholar
, Swain, M.Hinkel, E. (2005).The output hypothesis: Theory and research.Handbook on research in second language teaching and learning.Mahwah, NJ:, Erlbaum471-483 Google Scholar
Swain, M., Lapkin, S. (1995).Problems in output and the cognitive processes they generate: A step towards second language learning.Applied Linguistics.16,371-391doi:10.1093/applin/16.3.371 Google Scholar
VanPatten, B. (1990).Attending to form and content in the input.Studies in Second Language Acquisition.12,287-301doi:10.1017/S0272263100009177 Google Scholar
VanPatten, B. (1996).Input processing and grammar instruction in second language acquisition..Norwood, NJ:Ablex Google Scholar
Wilkins, D. (1976).Notional syllabuses..Oxford:Oxford University Press Google Scholar
Williams, J. (1999).Learner-generated attention to form.Language Learning.49,583-625doi:10.1111/0023-8333.00103 Google Scholar
Williams, J. (2001).The effectiveness of spontaneous attention to form.System.29,325-340doi:10.1016/S0346-251X(01)00022-7 Google Scholar
, Williams, J.Hinkel, E. (2005).Form-focused instruction.Handbook on research in second language teaching and learning.Mahwah, NJ:, Erlbaum673-691 Google Scholar
Zhao, S., Bitchener, J. (2007).Incidental focus on form in teacher–learner and learner–learner interactions.System.35,431-447doi:10.1016/j.system.2007.04.004 Google Scholar