Research in feedback has not found any one approach most affective for providing feedback on second language (L2) writing. In this article I will review some of the research on feedback that spans the past thirty-five years. Though much of the research has found that no approach seems to make a statistically significant difference in the student’s acquisition process, some textbooks for the training of foreign language teachers continue to advocate for one method over another.
I began carrying out a long term project at Montclair State University in the spring of 1994 where different forms of feedback were applied to higher-intermediate level ESL students of varying ability levels. Since 1995, I have also initiated similar research with students at Bergen Community College. To gain a better understanding of the forms of feedback in use at other institutions, I decided to survey fifty instructors of English to speakers of other languages. In the second part of this article I will discuss their responses to questions concerning feedback preferences.
Terms: Errors and Feedback
Before reviewing the research in this area, it is necessary to define two terms: errors and feedback. Our understanding of the word error itself can explain much of the disagreement that is to be found regarding its treatment. From a behaviorist viewpoint, errors were seen as “bad habits” which needed to be overcome through learning. For contrastive analysis, the counter part of bad habits was negative transfer. When elements from the student’s L1 differed greatly with structures in the L2, the possibility of interference was seen as great (Larsen-Freeman & Long 1991, 52-53).
With the Chomskyan inspired view of acquisition as rule driven, errors came to be seen as indicators of elements not yet fully acquired. Put simply, “errors” presented a failure in linguistic competence while “mistakes” were viewed as lapses in performance (Corder 1984, 24-25).
For the L2 student, “errors” most often represented an inadequate knowledge of rules. With the study of interlanguage, errors came to be interpreted as dialectal. In this continuum moving from the L1 to the target language, a student’s evolving interlanguage seemed to follow a “built-in syllabus” that, in the absence of fossilization, would lead to something resembling the L2 (Corder 1978, 74-77, 83).
In the seventies, Burt & Kiparsky (1974) distinguished global errors from local errors. Global errors referred to confusion in the sentence’s meaning caused by errors involving the major constituents of the sentence. One example of a global error in a complex sentence might be the incorrect use of a conjunction between clauses. In a simple sentence, a global error might refer to word order for example. Local errors on the other hand were errors found within a constituent, be it a clause in a complex sentence or a noun phrase, for example, in a simple sentence. It was suggested that perhaps students should first learn to correct global errors while instructors tolerate local errors.
With more communicative views of language acquisition, errors were recognized as listener defined. With this in mind, some may claim that only those elements that cause confusion on the part of the listener warrant correction.
The treatment of errors or the teacher response to errors is more commonly referred to now as feedback. Kulhavey in 1977 defined feedback as “any of the numerous procedures that are used to tell a learner if an instructional response is right or wrong” (Kulhavey 1977, 211).
The term has come to include all reactions that an instructor or anyone else (including the student him- or herself) might offer with reference to student production (or instructor performance) (Rinvolucri 1994, 287-288). Levels of intensity of feedback forms vary from meticulous corrections and suggestions to no feedback at all. In between these two extremes one finds feedback sensitive to particular pedagogical criteria and feedback with successful communication in mind.
Research Experiments
With such a wide spectrum of feedback forms, how does the instructor decide which to use? Certainly one approach must be more affective than another. A number of quantitative experiments have been carried out on L1, L2, and foreign language students. Some of the earliest experiments on feedback were carried out on L1 students.
Page in 1958 interpreting data from 2,139 secondary students claimed that those who received comments on their papers along with their score improved at a greater rate than those with just a grade and no comment. However it should be noted that Page purposely did not attempt to control many of the variables with one clear exception, treatment types (Page 1958, 174).
Another experiment involving L1 students was published in 1967 by Stiff. Treatments included marginal comments, summary comments and a mixture of both. A statistically significant difference was not discovered.
Several experiments have been carried out with the help of ESL or EFL students. Hendrickson (1981) carried out research to determine if local errors should be ignored by instructors. Correcting all local and global errors in one group and global only in the other, Hendrickson did not find a statistically significant difference in between the two parties. Robb, Ross, and Shortreed (1986) compared four types of feedback. They claimed that more direct forms of corrective treatment do not improve student writing at a rate greater than that of other methods. Several other experiments have up held this claim. In carrying out a short-term experiment on 72 students at Montclair State University, I also found no statistically significant difference between those who received teacher editing, circling of errors, and no comment feedback (Ihde 1993).
Two foreign language researchers, however, have found statistically significant results regarding differences between treatment groups. Lalande (1982) found that there was a statistically significant difference between post-test data of students receiving symbol and editing feedback. Those who received the symbol treatment had less grammatical and orthographic errors than their counterparts.
Students were required to rewrite their essays making use of the teachers’ feedback; however, students receiving symbol feedback were required to keep track of their errors from essay to essay. In 1984, Semke using four treatment groups reported that there was no statistically significant difference in the post-test results concerning writing accuracy; however, the test of writing fluency and the test of language proficiency both showed that comments on their own were superior to editing, comments with editing, and symbols. Semke also reported that approaches that force students to identify errors and correct them independently may have a negative effect on student progress. Students receiving symbol feedback and being required to revise their work performed significantly lower on the test of writing fluency.
More recent research projects including Kepner 1991, Carroll, Swain, and Roberge 1992, and DeKeyser 1993 have continued the search for data indicating efficacy of one approach over another.
Survey
With such differing results from the above cited research projects, it is not surprising to find similar diversity among practitioners. In the summer of 1993, I had the opportunity to elicit data from about forty French EFL teachers who were attending a summer course at Trinity College Dublin. I decided to ask them which feedback approach they used most often and why? Twenty-eight of these teachers responded. I later put the same questionnaire on the TESL-L list. I received twenty-two responses of which twenty-one were from Americans.
Before comparing responses received from French and American English teachers, let us become clear on what the different approaches entail. Perhaps the most meticulous of approaches is what Robb, Ross, and Shortreed called “correction” (1986, 86) and Stiff termed ”marginal feedback” (1967, 67-68). With this approach the instructor is actually editing lexical, syntactic, and stylistic errors and for this reason I have chosen to call it editing in this article. Rewrites for this method are, at best, exercises in copying. Advocates for this approach such as Lalande (1982, 140) and others (see Omaggio 1986, 50) see it as the only way for obtaining near-native fluency.
The use of corrective symbols is perhaps one of the more popular approaches. Labeled “symbolic code” by Semke (1984, 196), “coded” feedback by Robb, Ross, & Shortreed (1986, 86), and “direct correction treatment” by Hendrickson (1980, 218), this method identifies the place and type of error while not actually providing the correct usage.
“Terminal” feedback as used by Stiff (1967, 69-71) refers to an approach in which both symbols and comments at the end of the essay are used.
On the other hand, marking the place of the error without identifying the type has been termed “uncoded” feedback by Robb, Ross, & Shortreed and “indirect error treatment” by Hendrickson. The assumption here is that students will be able to figure the type of error committed. This approach will be referred to as circling here.
More communicative approaches in responding to written errors stress understanding. If the error causes confusion (see Burt 1975 and Semke 1984) or intolerance of any kind (see Ensz 1982 and Guntermann 1978) on the part of the reader, then its occurrence must be addressed. This is often done by writing the student a note at the end of the essay. This note could be in response to the subject of discussion as well as noting some errors. I will term this as summary.
Results
When comparing use of feedback types between my French and American informants, clear differences could be seen. No one approach was shared by a majority of Americans. Thirty-eight percent of the American sample claimed to make use of circling, thirty-three percent maintained that they used symbols, and twenty-four and five percent claimed to use editing and summary techniques respectively. In contrast to this, nearly three-quarters of the French sample made use of one method alone, symbols. Circling and editing accounted for only eighteen and eleven percent of the respondents. No French participant claimed to employ summary methods. The French EFL teachers supported their choice of symbols as the preferred approach by claiming that it caused students to become more aware of the different types of errors being committed. Many also claimed that symbols made students think for themselves and aid in developing self correction skills.
A less direct approach than this, circling, was used by five out of twenty-eight of the French instructors and eight out of twenty-one of the American instructors. American teachers in defending their approach maintained that responsibility needs to be placed on the student for identifying the errors. Some instructors stated that it is often unclear what the cause of the error is. This approach permits the student to improve on what they originally wanted to say and not what the instructor perceived them as wanting to say. Lastly one professor claimed that if the element is really a mistake and not an error, the student is well capable of correcting it without teacher intervention.
The French instructors who made use of circling wanted their students to discover what was wrong with their errors. This, one of the teachers maintained, is the job of a student and not the instructor. Another respondent claimed that classifying each type of error with symbols would be too time consuming, especially seeing that the most common errors are regularly reviewed in class.
None of the participants in the survey, American or French, claimed not to make any corrections on students’ essays. In that aspect all agreed that some form of teacher feedback was necessary. As concerned the use of instructors editing students’ papers, nearly a quarter of the Americans used this approach whereas only about eleven percent of the French sample used such practices. One French EFL teacher stated that this method provided personalized help for students and it allowed them to reflect on their errors outside of the classroom as well.
As stated above a larger percentage of American ESL instructors used full editing. Though most did this by writing on the students’ papers, one respondent made use of a tape recorder to inform students of their errors. The reasons for using editing were varied. One instructor claimed his students deserved such correction after all the hard work they put in on writing their essays. Another stated that students were sometimes confused by symbols and unable to correct the errors when they did understand the symbol.
Participants were also asked about their use of rewrites and the frequency of them. Results showed that whereas all American informants claimed to ask their students to rewrite their work, twenty-two percent of the French sample did not. The frequency at which instructors did request rewrites varied. Fifty-seven percent of the American ESL teachers maintained that they always had students work on essays in several drafts. Twenty-nine and fourteen percent of the same sample claimed “most of the time” and “sometimes” respectively. Though three of the French respondents did not mark frequency, the only category to register over eleven percent was “sometimes” which reached forty-six percent.
One reason for the large number of “always” responses in the American sample may be due to the effects of process writing (see White & Arndt 1990). Part of this approach is based on the concept of several drafts. One American respondent made an interesting comment which may shed light on this discussion. She claimed that “always” was a viable option because her students did their work on word-processors.
Due to the small sample sizes and the lack of random selection of participants, one cannot generalize these findings. However as a preliminary study, one can conclude that three points stand out. First, the instructors in the sample agreed that some form of correction is necessary. Secondly, the use of summary methods, which were identified above as possibly being communicative in nature, were not used by most teachers in the sample. Lastly, whereas most instructors disagreed on the frequency of rewrites, a small number of instructors (all of whom were French) did not use rewrites at all.
These data validated the use of certain feedback treatment forms for the long term project which I spoke of at the beginning of this article. Symbols have been applied in correcting essays of one of the treatment groups. This seems to be one of the commonly used methods of correction as can be seen in the above data. The other treatment group in the long term experiment received summaries at the end of their essays. This was used as the opposing treatment due to the above data which clearly shows the infrequency with which teachers use this approach.
References
Burt, M. K. & C. Kiparsky. 1974. “Global and local mistakes,” in J. Schumann and N. Stenson (eds.). New Frontiers in Second Language Learning. Rowley, MA. Newbury House, 71-80.
Burt, M. K. 1975. “Error analysis in the adult EFL classroom.”TESOL Quarterly 9, 53-63.
Carroll, S., M. Swain, & Y. Roberge, 1992. “The role of feedback in adult second language acquisition: error correction and morphological generalizations.” Applied Psycholinguistics. 13.2, 173-198.
Corder, S. P. 1978. “Language-learner language,” in J. P. Richards (ed.) Understanding Second and Foreign Language Learning: Issues and Approaches. Rowley, MA: Newbury House., 71-93.
Corder, S. P. 1984. “The significance of learners’ errors,” (IRAL. 5.4, 1967) reprinted in J. C. Richards (ed.)Error Analysis: Perspectives on Second Language Acquisition. London. Longman. 19-27.
DeKeyser, R. M., 1993. “The effects of error correction on L2 grammar knowledge and oral proficiency”. Modern Language Journal. 77.4, 501-514.
Ensz, K. Y. 1982. “French attitudes toward typical speech errors of American speakers of French.” Modern Language Journal. 66, 133-139.
Guntermann, G. 1978. “A study of the frequency and communicative effects of errors in Spanish.” Modern Language Journal. 62, 249-253.
Hendrickson. J. M. 1980. “Treatment of error in written work.” Modern Language Journal 64:2, 216-221.
Hendrickson. J. M. 1981. Error Analysis and Error Correction in Language Teaching (Occasional Papers No. 10). Singapore: SEAMEO.
Ihde, T. W. 1993. “Error correction in L1 & L2 language learning.” Rule Syntactica 4. 46-53. (ERIC Document Reproduction Service Number ED 374 678).
Kepner, C. G. 1991. “An experiment in the relationship of types of written feedback to the development of second-language writing skills.” Modern Language Journal .75, 305-313.
Kulhavey, R. W. 1977. “Feedback in written instruction.” Review of Educational Research. 47, 211-229.
Lalande, J. 1982. “Reducing composition errors: an experiment.” Modern Language Journal. 66, 140-149.
Larsen-Freeman, D. and M. H. Long. 1991. An Introduction to Second Language Acquisition Research. New York. Longman.
Omaggio, A. C. 1986. Teaching Language in Context: Proficiency-Oriented Instruction. Boston. Heinle and Heinle Publishers.
Page, E. B. 1958. “Teacher comments and student performance: a seventy-four classroom experiment in school motivation.” Journal of Educational Psychology. 49, 173-181.
Rinvolucri, M. 1994. “Key concepts in ELT: feedback,” ELT Journal. 48.3, 287-288.
Robb, T., S. Ross, & I. Shortreed. 1986. “Salience of feedback on error and its effect on EFL writing quality,” TESOL Quarterly. 20.1, 83-93.
Semke, H. D. 1984. “Effects of the red pen,” Foreign Language Annals. 17, 195-202.
Stiff, R. 1967. “The effect upon student composition of particular correction techniques.” Research in the Teaching of English 1. 54-75.
White, R. & V. Arndt. 1991.Process Writing. New York: Longman.
