Abstract
The preceding chapter has adopted a predominantly pedagogic perspective by discussing the possible effects of oral and written error correction with respect to the development of explicit and implicit knowledge, the distinctive characteristics of the two types of feedback, as well as the decisions that teachers have at their disposal in this respect, offering simultaneously some comments on the value of specific corrective techniques.
This is a preview of subscription content, log in via an institution to check access.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
Notes
- 1.
Ellis (2010b) draws attentionto the fact that different theoretical justifications can be provided for each of these two distinctions. In the case of input-providing and output-inducing corrective feedback, for example, the claims of cognitive interactionist theoretical positions, such as the modified version of the Interaction Hypothesis (Long 1996), according to which it is input that constitutes the driving force of second language development, are pitted against the tenets of Skill-Learning Theory (De Keyser 1998, 2001), which posits that the production of output is indispensible for acquisition.
- 2.
Truth be told, the distinction between fluency-oriented and accuracy-based activities would be rather difficult to maintain in the present discussion on account of the fact that the types of interaction during which CF was provided in particular studies are sometimes difficult to pinpoint and they in most cases fall somewhere in between relatively free communication and the performance of code-related activities.
- 3.
The direct consequence of the broader focus of research into incidental focus on form is reliance on a different unit of analysis than that typically employed in studies of oral corrective feedback. This is usually a focus on form episode (FFE), “(…) which includes all discourse pertaining to the specific linguistic structure that is the focus of attention” (Loewen 2003, p. 318). This allows researchers to investigate not only reactive (i.e. error correction) but also preemptive (i.e. before an error is made) focus on form.
- 4.
A brief description of this coding scheme, based on Allen et al. (1984), can be found in note 10 in Chap. 2.
- 5.
It should be pointed out that in the studies conducted by Lyster and Ranta (1997) and Lyster (1998b), an additional category of unsolicited uses of the first language was included, which, of course, cannot be treated as errors per se, but may be regarded and responded to as such by many teachers.
- 6.
Panova and Lyster (2002, p. 590) also included in their analysis “(…) a type of clarification request that focused on the literal, unintended meaning of learner utterance”. Even though this as well constitutes departure from the analytical framework used by Lyster in Ranta (1997), it does not entail the need to introduce an entirely new category.
- 7.
Tailor-made tests were used prior to that in what is known as text-reconstruction activities, such as those based on the idea of strategic interaction (DiPietro 1994) or those using the dictogloss procedure (Swain 1998).
- 8.
See footnote 3 earlier in this section for the definition of FFEs.
- 9.
An interesting reanalysis of these data as well as those procured in the course of earlier studies (e.g. Loewen 2004) is reported by Loewen (2007). Here, in addition to using the individualized test items from Loewen (2005), the researcher also investigated prior and subsequent use of the targeted forms in 4.5 h of classroom interaction with the help of a corpus analysis software program. He found no relationship between correct subsequent use of a specific feature and successful uptake as well as the results of individualized posttests, although he reported that, on the whole, the subjects were less accurate before the occurrence of form-focused episodes than afterwards. In contrast to the study outlined in the text, such findings clearly cast doubt on the significance of uptake as a measure of language learning. This led Loewen (2007, pp. 144–115) to comment that “(…) these findings suggest that studies of uptake should continue to be cautious in interpreting its significance. (…) [and] illustrate the importance of measuring learners’ L2 knowledge in a variety of ways”.
- 10.
For the sake of convenience, the term experimental is used here to refer both to true experiments, or studies conducted in laboratory settings, and quasi-experiments, or research projects carried out in real classrooms and using intact learner groups.
- 11.
A detailed description of the construction and scoring of these instruments as well as the others included in the battery is beyond the scope of the present chapter and can be found in the papers included in a recent publication edited by Ellis et al. (2009). The interested reader is also referred to the paper by Mystkowska-Wiertelak (2011), which offers an interesting and instructive critique of the use of oral elicited imitation as a measure of implicit knowledge.
- 12.
While the total scores provide information about the overall performance of the groups, thus making it possible to draw conclusions about the effectiveness of the two types of correction, the appearance of grammatical and ungrammatical items may reflect reliance on different types of knowledge (i.e. implicit vs. explicit), and the distinction between old and new items allows insights into the extent to which the improvement is the outcome of item or system learning (cf. Ellis 2005b, 2009b).
- 13.
The results concerning the performance of the whole groups are reported in an earlier paper by Ammar and Spada (2006), which is referred to in the following section as part of the discussion of research into the effects of input-providing and output-prompting oral error correction.
- 14.
The scale is a simplified version of the sequence of eight developmental stages originally proposed by Zobl (1984).
- 15.
The students were divided into the high-anxiety group and low-anxiety group taking into account the total mean score and standard deviation. More precisely, those who scored one standard deviation above the mean were considered to manifest high anxiety levels and those who scored one standard deviation below the mean were regarded as displaying low anxiety levels. The participants who scored between the two values were excluded from the analysis.
- 16.
As Dörnyei (2005, p. 57) writes, the working memory span is “(…) a robust predictor of a wide range of complex cognitive skills and it is highly correlated with performance on the type of reasoning tasks that underpin standard tests of intelligence”. The test is used to measure processing and storage of information in a dynamic and simultaneous way.
- 17.
Ellis (2007) discusses the difficulty involved in the acquisition of these two features with respect to such criteria as grammatical domain, input frequency, learnability, explicit knowledge, reliability, scope, formal semantic redundancy, and experts’ opinions.
- 18.
The feedback was provided in exactly the same way as in the study conducted by Ellis et al. (2006) reported above. The testing instruments were also identical.
- 19.
It should be noted that Mackey’s (2006b) suggestions are made in relation to interactionist research in its entirety rather than only studies of oral corrective feedback.
- 20.
There is no agreement among researchers as to the value of learners’ incorporating into their own production the reformulation contained in a recast. On the one hand, researchers such as Mackey and Philp (1998) take the stance that whether or not learners actually manage to repeat a recast may be inconsequential to acquisition and it is redundant in the error correction sequence that is both initiated and completed by the teacher. On the other hand, however, according to the tenets of the Output Hypothesis (Swain 1985, 1995, 2002), even such production can assist the move from semantic to syntactic processing, aid the process of hypothesis-testing, and contribute to greater automatization.
- 21.
A very different evaluation of the efficacy of recasts can be found in the paper by Goo and Mackey (2013), who point to a number of methodological flaws and interpretative problems in the studies conducted to date.
- 22.
- 23.
Samuda’s (2001) study in fact consisted of three stages, in which the learners first performed a communicative task without any intervention on the part of the teacher, then they made presentations to the whole class during which feedback was provided, and, finally, they were requested to prepare a poster on their own, with the teacher yet again adopting the non-directive role of an observer.
- 24.
The six subtypes of reformulation were as follows: (1) isolated recast minus prompt, (2) isolated recast plus prompt, (3) embedded recast minus prompt, (4) embedded recast plus prompt, (5) recast plus enhanced prompts, and (6) recast plus expansion. The five subtypes of elicitations included: (1) unmarked elicitation, (2) marked elicitation, (3) marked elicitation plus prompts, (4) marked elicitation plus enhanced prompt, and (5) elliptical elicitation. He also isolated the category of other feedback that contained explicit correction, repetition with falling intonation and various content negotiation moves.
- 25.
It should be kept in mind, however, that the study focused both on negotiation of form and meaning, with the effect that it did not attempt to tease apart the differential effects of confirmation checks and clarification requests following utterances that were genuinely misunderstood and those where the negotiation move was meant to be corrective in nature. On the whole, the incidence of negotiated interaction was rather low (0.66 such exchanges per task), it was predominantly conversational rather than didactic in nature, and the adjustments made by interlocutors in response to feedback were in most cases minimal. Interestingly, similar findings have been reported by Pica (2002), who analyzed discussion activities in L2 content classes. To be more specific, the incidence of interactional feedback in the form of recasts and negotiation moves was low, with the effect that output modifications hardly ever occurred. This led the researcher to conclude that such activities fail to provide learners with both positive and negative evidence, the latter of which is particularly significant for language learning.
- 26.
In fact, the considerable improvement on the delayed posttest could be attributed to the fact that the learner took part in stimulated recall sessions in the interval between the two tests which involved commenting on the corrective episodes. Thus, the evidence for the positive impact of recasts becomes somewhat tenuous.
- 27.
In fact, this applies in equal measure to other types of corrective feedback that are often compared with recasts, an issue that will be touched upon later in the present section.
- 28.
An interesting discussion of the methodological issues involved in the coding of corrective recasts can be found in Hauser (2005), who warns that coding schemes ignore the construction of meaning by participants in the local context of interaction.
- 29.
It should be noted that the errors were also corrected in the written versions of the report by means of circling and juxtaposing them with reformulations. Given the availability of the oral feedback to all learners, however, it appears justified to discuss the study in this section rather than the one dealing with written error correction.
- 30.
In both this study and the one undertaken by McDonough and Mackey (2006), the developmental levels were established in accordance with the scale developed by Pienemann and Johnston (1987).
- 31.
Syntactic priming, also known as structural priming, is defined as the use of a structure that has been previously heard or spoken in subsequent utterances (Bock 1995). Two experiments investigating its occurrence in interactions between L2 speakers of English are reported by McDonough (2006).
- 32.
Although Ishida (2004) did not include a control group that would have only taken part in negotiated interaction, the time-series design allowed her to document the progress as a result of the intervention, not only from the pretest to the posttest, but also from one instructional session to the next, with the effect that the subjects acted as their own controls. This is the reason why the study is discussed together with research projects actually comparing the effects of interaction with and without recasts.
- 33.
It should be noted that two students also participated in a delayed posttest conversational session that took place after 7 weeks.
- 34.
There have also been attempts to investigate the relative effects of explicit feedback and prompts, as evidenced by the research project by Nipaspong and Chinokul (2010), focusing on the development of pragmatic awareness. This is not an important line of inquiry, however, and one might wonder in fact whether it is at all possible to isolate the effects of the two given that prompting also involves the provision of highly explicit metalinguistic clues.
- 35.
It should be noted that the two distinctions are bound to overlap to some degree in studies seeking to explore the value of input-providing and output-pushing feedback options, such as those conducted by Lyster (2004) or Ammar and Spada (2006). This is because, although prompts may differ considerably in the degree of their explicitness, they are typically more overt than recasts, a somewhat extreme example being the provision of metalinguistic information.
- 36.
In fact, they equate the distinction between explicit and implicit feedback on errors in speech with that between direct and indirect feedback on written errors, which is highly problematic, as the learner is always cognizant of the corrective force of the indications included in a piece of writing, whether these provide the accurate forms or merely serve the purpose of highlighting the problems.
- 37.
It should be noted that Lyster and Saito (2010) did not consider the difference between explicit and implicit feedback as such, but looked at the effects of prompts, recasts and explicit correction, which renders the interpretation provided by the present author somewhat speculative. Still, elements of explicitness can be found both in metalinguistic feedback and elicitations, which are prompts, and direct correction, with the effect that more overt CF options can be regarded as more likely to foster the acquisition of the targeted language forms.
- 38.
Worth mentioning is also the study conducted by Adams, Nuevo and Egi (2012), which generated evidence for the link between the use of implicit CF in the form of recasts, output modifications, and gains in explicit knowledge, but failed to find such an advantage for explicit CF in the form of direct correction. It is not considered here in detail, however, because it examined peer correction during learner-learner interactions while the focus of the present section is on expert correction, whether delivered by the teacher, native speaker, or via the computer.
- 39.
It is worth pointing out that Goo and Mackey (2013) are rather skeptical of the value of such studies, arguing that they suffer from methodological flaws, related, for example, to the failure to control for modified output opportunities, the comparison of a single variable with multiple variables, the presence of form-focused instruction, as well as unclear contributions of prior knowledge and out-of-class exposure. In the opinion of Lyster and Ranta (2013, p. 181), however, these concerns are overstated, for the reason that applied SLA researchers should be “(…) concerned with investigating SLA phenomena that are of practical significance to teaching and with conducting research in such a way that it is transparently relevant to teachers”.
- 40.
It should be explained that proficiency was defined here as the participants’ mastery of the targeted features on the pretest rather than in general terms.
- 41.
The contrasting results of the studies conducted by Sheen (2007a) and Goo (2012) are likely to stem from the fact that they operationalized and measured aptitude in different ways. Another possible explanation is that they involved different instructional targets.
- 42.
Indeed, it is possible to view developmental readiness as both an attribute of the learner, as he or she has to be psycholinguistically ready to internalize a particular structure or capable of performing the requisite processing operations, or as a property of that structure, since some linguistic features may be developmentally early and others late, and there are also such that are variational in nature, i.e. they are not constrained by developmental stages or processing operations (Pienemann and Johnston 1986). The decision to regard developmental readiness as a linguistic factor reflects the way in which it was classified in Sect. 4.2.2, where the framework for investigation error correction was introduced.
- 43.
Such effects, however, were less clearly visible for the other two targeted forms, that is plurals and past tense, which shows that learners’ cognitive response interacts with linguistic factors (Egi 2007).
- 44.
It should be stressed one more time that the present author fully concurs with Sheen (2010c) and Sheen and Ellis (2011), who make the point that written corrective feedback can only be explicit. What Ferris (2010) seems to have in mind in this quote in fact is the distinction between direct and indirect feedback, which could, in everyday parlance, be viewed as differing with reference to the degree of their explicitness.
- 45.
On account of the fact that the study had several foci, it also serves as an illustration of how learner engagement with written corrective feedback can be investigated.
- 46.
In fact, the procedure involved the learners not only reading the story before being asked to rewrite it but also the teacher reading it aloud so that the students could jot down the key words, which makes the activity similar to the dictogloss, a text-reconstruction task frequently employed by researchers in the area of form-focused instruction (e.g. Swain 1998).
- 47.
Language-related episodes are to a large extent identical to focus-on-form episodes defined in note 3 earlier in this chapter, but the former rather than the latter tend to be used when analyzing interactions between learners that take place in pairs or small groups.
- 48.
These weaknesses are most often related to the lack of a true control group, the failure to control for all the extraneous variables, the nature of the outcome measures and assessment procedures, and the presence of only one posttest, which precludes the researchers from advancing claims about the long-term contributions of different types of treatment.
- 49.
According to the classification introduced in Sect. 3.5.4.2 in Chapter 3, (3) is also an example of indirect feedback since the correct version is not provided by the teacher. Bitchener and Knoch (2010), however, view it as a form of direct feedback.
- 50.
In most of these studies, the main emphasis is laid on the quality of learners’ noticing of the changes made to their initial texts. Since noticing is reflective of learner response to feedback, the discussion here is only confined to the impact of reformulation on subsequent revisions.
Author information
Authors and Affiliations
Corresponding author
Rights and permissions
Copyright information
© 2014 Springer-Verlag Berlin Heidelberg
About this chapter
Cite this chapter
Pawlak, M. (2014). Research on Error Correction. In: Error Correction in the Foreign Language Classroom. Second Language Learning and Teaching. Springer, Berlin, Heidelberg. https://doi.org/10.1007/978-3-642-38436-3_4
Download citation
DOI: https://doi.org/10.1007/978-3-642-38436-3_4
Published:
Publisher Name: Springer, Berlin, Heidelberg
Print ISBN: 978-3-642-38435-6
Online ISBN: 978-3-642-38436-3
eBook Packages: Humanities, Social Sciences and LawEducation (R0)