«Item type text; Electronic Dissertation Authors Goertler, Senta Publisher The University of Arizona. Rights Copyright © is held by the author. ...»
(1) Explicit correction: teacher supplies the correct form and clearly indicates that what he student has said was incorrect; (2) Recast: teacher implicitly reformulates all or part of the student’s utterance; (3) Elicitation: teacher directly elicits a reformulation form students by asking question such as “comment ça s’appelle?” or “How do we say that in French?” or by pausing to allow students to complete teacher’s utterance, or by asking students to reformulate their utterance; (4) Metalinguistic clues: teacher provides comments, information, or questions related to the well-formedness of the student’s utterance such as “ça ne se dit pas en francais” or ‘C’est masculin?”; (5) Clarification request: teacher uses phrases such as “Pardon?” and “I don’t understand”; (6) Repetition: teacher repeats the student’s ill-formed utterance, adjusting intonation to highlight the error (p. 4).
Looking at corrective feedback types in general, several studies compared corrective feedback types and their effectiveness in SLA, as well as the relationship between error type and feedback type. In regard to implicit and explicit feedback, Morris (2002) found 70% to be implicit, while Seedhouse (1997) confirmed a low percentage of explicit feedback. In conversations only little explicit feedback is expressed (Chun, Day, Chenoweth, & Luppescu, 1982). Williams and Evans (1998), however, argued that the explicit treatment of structures would be effective for simple forms for which the learners were ready. However, explicit feedback seems to be rare in today’s classrooms.
Seedhouse (1997) argues that the absence of such explicit feedback may make it harder for learners to perceive the feedback.
During my review of studies performed in the field, it became apparent that corrective feedback is confusing at the least. While the definitions offered appear to be clear, actual data categorization is not always as clear. Two of the issues in the review of the literature are that (1) applications of the definitions differed and (2) applications of the definitions were not always provided. Especially the division of negative feedback into implicit and explicit feedback seems troublesome. Any form of implicit feedback can be used in an explicit way by using intonation, gestures, or written tools. For example, a teacher can repeat a student’s utterance and then write the incorrect form on the board, maybe even circling the error in red. While most people would classify such a correction as explicit, if only the incorrect form is written on the board in addition to the verbal repetition the definition becomes more challenging.
Within the options of implicit feedback, it appears that forms that engage the learner more actively, such as negotiation moves, are more successful. Doughty (1994, as cited in for example Nicholas, Spada, and Lightbown, 2001) reported that 61 out of the 284 recasts, i.e., 21%, resulted in learner repetition, here considered as evidence of noticing, with no difference between recast with or without emphasis. In this case, one must consider whether a recast with emphasis is still implicit feedback. Also, a mere four of the 173 repetitions resulted in learner repetition. Doughty saw these findings as evidence that learners are able to perceive the difference between a repetition and a recast, although she is cautious about the effectiveness of recasts due to the low repetition rate. Similarly, Lyster and Ranta (1997) questioned the effectiveness of recasts, since in their study these forms had the lowest uptake rate. Morris (2002) found an overall low repair rate, yet did find negotiation moves more effective than recasts. Lyster and Ranta (1997) found recasts to have the lowest, and elicitation and metalinguistic feedback to have the highest rate of uptake (Lyster & Ranta, 1997). Lyster and Ranta measured effectiveness of error correction by the opportunity to negotiate forms, and hence considered metalinguistic feedback, elicitation, clarification request, and repetition of errors beneficial, whereas recasts and explicit correction were not. Lyster (1998a) triangulated the results in a later study, and found that the corrective feedback forms that promote negotiation of forms resulted in more immediate repair than those that did not.
This claim is based on the assumption that if learners have to actively work on retrieving the correct utterance, they are more likely to remember it. It appears that it is easier for learners to notice forms that they were pushed to self-repair, rather than those simply provided by the teacher.
After this general discussion of effectiveness, one must also consider the relationship between the error type and the corrective feedback type. Studies investigating this relationship suggested that speakers familiar with NNSs create patterns of correction. Morris (2002) analyzed peer feedback in dyad work in a beginning Spanish class and found that morphosyntactic errors were followed by recasts and lexical errors by negotiation moves. Lyster (1998a) found that recasts were used in response to grammatical and phonological errors, while negotiations of form, such as elicitation, metalinguistic feedback, clarification requests, and repetitions, were used for lexical errors. Mackey, Gass and McDonough (2000) investigated the types of feedback and their correlation with error type in ESL and Italian-as-a-foreign-language (IFL) dyads.
Recasts were found to be the preferred form of corrective feedback for morphosyntactic errors (i.e., 75%). Phonological errors on the other hand received negotiation moves, especially clarification requests, as feedback in the ESL group, but recasts in the IFL group. Learners were able to perceive the feedback on phonology and lexis as evidenced in stimulated recall. However, they were unlikely to perceive the feedback on morphosyntactical errors, further suggesting that recasts are difficult to perceive. There appears to be a relationship between the error type and the feedback type. It is, however, not clear whether the dominant feedback type used in response to a certain error type is the most effective feedback type for such errors.
The largest concern with corrective feedback is that learners do not repair all mistakes after corrective feedback (Day, Chenoweth, Chun, & Luppescu, 1984; Lyster, 1998; Lyster & Ranta, 1997; Oliver, 1995, 2000) and may not even recognize corrective feedback as such (Mackey, Gass, & McDonough, 2000). This means that it is impossible to use only one measure of effectiveness to determine the outcome of certain corrective feedback moves. Yet, the following questions remain and cause problems for research design: (1) If a learner does not show evidence of recognizing that he/she was corrected, does that always mean that he/she in fact did not recognize the correction? (2) Even if a learner does not recognize corrective feedback as such consciously, does this necessarily mean that it can never lead to language learning? (3) Even if there is evidence of uptake, does this mean that as a consequence the learner’s language has improved or will improve?
The trouble with research on corrective feedback in SLA is that researchers do not seem to agree on the terminology or methodology, hence it becomes difficult to make any claims that the majority of researchers will accept. In reviewing the literature, it is apparent that more research needs to be done with a clear definition of the corrective feedback, measuring development over time, as well as noticing in various forms such as private speech, stimulated recall, and uptake (in Lyster and Ranta’s definition, 1997).
2.3.4 Gaps in the Literature While corrective feedback has been studied extensively in face-to-face (F2F) conversation, it has not been given much consideration in the CMC environment. Most research has been conducted in either ESL or immersion contexts, creating the necessity for more studies investigating the role of corrective feedback in the FL classroom, especially in a CMC context. Further studies are also needed on dyads between NNSs, as these have been less common than studies of dyads of NNSs and NSs. Finally, variations in findings and outcomes among studies suggests the need for further systematic and detailed research on corrective feedback including detailed descriptions of applied settings, methods, and terminology.
2.4. CMC As described in chapter one, CMC is communication mediated through the computer and can take asynchronous (Email, Message Boards) as well as synchronous forms (Local Area Networtk (LAN)/Intert Relacy Chat (IRC) chats, Multiple User Dimension (MUD), Object Oriented (MOO)s). In practical and research applications it has been used both within in the same physical space and also through distant learning set-ups. CMC forms share the following features: anonymity, written form of communication, slower than spoken, can be used spontanteously, editing capabilities, options to re-read messages, and the option to post messages simultaneously.
Furthermore, as a written medium, pronunciation is not an issue. Although CMC is written, CMC, especially synchronous CMC, shares many aspects with the oral genre of communication such as a conversational style, spontaneity, and informality (e.g.,.
Böhlke, 2003; Gonzalez-Bueno, 1998; Kern, 1995). Yates (1996) argues that CMC is closer to written than the oral modality. CMC shares features of both a typical written and a typical oral form of communication (Chun, 1994; Kern, 1995; Warschauer, 1996).
CMC has been argued to be a transitional step between the two (Chun, 1994).
In the absence of paralinguistic features, everything has to be expressed in written words (Donaldson & Kötter, 1998). Over the course of the years, separate conventions to make up for the lack of paralinguistic signs have been developed, such as emoticons, all CAPS, or extensive punctuation (Kern, 1995). The advantage of computer-mediated communication is also that emphasis can be portrayed visually and without distractions of gestures and intonation.
Most of the early research and use of CMC in the classroom was utilizing the Daedalus Integrated Writing Environment’s synchronous discussion feature: InterChange (see Kern, 1995; Ortega, 1997). Today, language teachers can choose a variety of different software programs affording them the opportunity for remote as well as local CMC, both synchronous as well as asynchronous, and also with or without picture, video, and voice capabilities.
2.4.1 Problems in Research Design:
A review of previous CMC research sheds light on several investigative challenges. Research design and theoretical grounding were problematic, especially in early CMC research. Early studies compared the benefits of using CMC with the benefits of classroom instruction, although some studies did not even provide comparison values.
Other studies simply did not compare similar enough contexts to be considered valid evidence, such as a teacher-fronted regular class with CMC pair work. In addition, as with any language learning activity, the confoudning variables are numerous and diverse.
Most studies, however, either ignored or failed to report variables that may have had an impact on findings, such as activity design, task type, learner profiles, group size, and lab set-up. Many studies only consulted the transcripts as data for their study, which only gives a partial picture of the actual interactions and learning occurring in online chatting.
As Smith (presentation at CALICO, 2005) has indicated, for CMC research to be valid, information about what happened in the physical environment must be considered.
Although Smith focused specifically on chatting in his study, his methodological recommendation was made for all forms of CMC research.
One of the greatest flaws in CMC research has been the lack of comparability, due to the lack of information about details of the studies, such as factors that influence learner interaction. The following section describes some of these factors.
2.4.2 CMC and Language Learning Operating within an interactionist framework, the mere fact that learners have the opportunity to process input, produce output, negotiate meaning, notice, and receive conversational feedback would suggest that CMC is beneficial for learning. However, due to the nature of the communication medium, such conclusions cannot easily be drawn without empirical evidence.
Most of the earlier studies in CMC compared classroom face-to-face interaction with CMC contexts, and investigated the effects each had on various aspects of language learning. Beauvois (1992, 1997, 1998), Chun (1994) and Kelm (1992) all argued that CMC encourages more target language use. The accountability, i.e., the fact that the student is faced with his or her own lanauge production on the computer screen, the presence of transcripts as evidence of conversation, and the slower speed of the interaction may all be considered contributing factors for the increase in target language use. Ene et al., although not comparing CMC with face to face (F2F) interaction, also found a high percentage of target language use.
Kelm (1992) argued that CMC improves reading comprehension. Since CMC is a written medium of conversation even when used synchronously, this argument stands to reason. The same logic, then, can be applied for writing ability. Chun (1994) found that CMC furthered the writing ability of FL learners, and Donaldson and Kötter (1999) argued that the absence of paralinguistic features and the consequential need to express everything in written words would benefit language development.
Many of the enthusiastic early researchers in CMC argued that CMC can benefit language development in general (Beauvois, 1992; Chun, 1994; Kelm, 1992; Kern, 1995;
Warschauer, 1996). However, caution was expressed in regards to complexity of language production (Chun, 1994) and accuracy of postings in CMC (Beauvois, 1992;
Kern, 1995). Gonzalez-Bueno (1998) found that the CMC learner language in an Email context exhibited higher quality of language in terms of accuracy, but in a later study, Gonzalez-Bueno and Perez (2000) found that the electronic journals had fewer lexical but more grammatical errors than the paper-and-pencil entries. Abrams (2003) found synchronous CMC beneficial for increased language quantity, but not necessarily quality.