Composition Forum 45, Fall 2020
http://compositionforum.com/issue/45/

Is Feedback on Grammar Harmful or Helpful? Questionable Answers and Unanswered Questions

Kristen di Gennaro and Monika Ekiert

Abstract: Current composition practice relies on a decades-old summary of research concluding that a focus on grammar in students’ writing is useless, or even harmful. Conversely, hundreds of recent studies from the fields of second-language writing and applied linguistics claim to provide evidence of the benefits to providing feedback on grammar in students’ writing. This article summarizes the arguments for and against such feedback and problematizes the results of previous research by describing a quasi-experimental study measuring the effects, both positive and negative, of providing students with grammar feedback on their writing. Results show that, while feedback on specific grammatical forms improved participants’ accuracy on those forms, it also led to decreased accuracy on other forms related to but not the focus of instruction. Furthermore, the control group’s accuracy equaled or surpassed that of the two feedback groups.

Introduction

Current post-secondary writing pedagogy recommends incorporating minimal, if any, formal grammar instruction into writing classes, a result, in part, of research conducted decades ago. Ever since the 1963 report in which Richard Braddock, Richard Lloyd-Jones, and Lowell Schoer summarized research on the (lack of) effectiveness of grammar instruction on students’ written compositions, scholars have cited and agreed with their finding that grammar instruction is unnecessary or even harmful for students in writing classes (see Susan Peck MacDonald for a critique of the Braddock et al. report and its impact on writing teachers’ knowledge). The subsequent dissemination and anthologizing of Patrick Hartwell’s 1985 essay, “Grammar, Grammars, and the Teaching of Grammar,” in which the author classifies and challenges the value of different types of grammar instruction, has contributed further to compositionists’ skepticism regarding the value of grammar instruction in writing classes. Most recently, such skepticism is deepened by the field’s current move away from a focus on prescriptive language rules in general, and toward instruction and assessment that take into consideration issues of social justice as witnessed by Vershawn Ashanti Young’s opposition to teachers enforcing Standard English policies in composition classes (Should Writers Use They Own English?) and Asao Inoue’s call for anti-racist pedagogies in writing instruction and assessment (How Do We Language?).

Whether or not writing instructors today dedicate class time to teaching grammar, many are still likely to provide feedback on students’ writing that calls some attention to grammatical accuracy, perhaps more so in composition courses taught by teachers coming from literary studies or creative writing rather than from composition and rhetoric programs. Even Richard H. Haswell’s frequently cited method of “minimal marking” attends to grammar by identifying, albeit without correcting, students’ grammatical inaccuracies in their writing (Haswell, Minimal Marking). And as Daniel Cole notes in What If the Earth Is Flat?, faculty teaching in Writing Across the Curriculum programs, who may also be unfamiliar with current composition pedagogy, have been found to focus a great deal (if not exclusively) on grammar and mechanics in their feedback on students’ writing. Arguments for attending to grammar in students’ writing may also refer to the changing demographics of the student body. Indeed, composition teachers assigned to teach multilingual classes might maintain that attention to grammar, at least in their feedback on students’ writing, is part of their mandate.

In short, while compositionists do not promote direct, in-class grammar instruction in mainstream post-secondary writing courses, drawing students’ attention to grammar in the context of feedback on writing assignments remains a common practice, especially in WAC and in courses with large numbers of multilingual students. And despite calls from compositionists to question practices stemming primarily from personal belief and collective lore in favor of practices based on evidence from research, as well as calls for replication studies in composition to support or reveal problems with earlier studies (see Elizabeth G. Allan and colleagues’ The Source of Our Ethos and Chris Anson’s The Intelligent Design of Writing Programs), recent research on the effectiveness, or not, of providing grammar feedback on composition students’ writing is scant.

Existing Research and Ongoing Debate

The apparent paucity of research in this area, despite its potential impact on practice, reflects what Haswell refers to as NCTE’s “war on scholarship,” that is, a dramatic decline in research that can be described as replicable, aggregable, and data-supported, published in composition’s flagship journals (NCTE/CCCC’s Recent War on Scholarship). The most relevant studies cited on this topic are primarily descriptive and not specifically about corrective feedback and its effectiveness, but rather analyses of students’ error patterns (e.g., Mina P. Shaugnessy’s Errors and Expectations and Andrea A. Lunsford and Karen J. Lunsford’s “Mistakes Are a Fact of Life), rankings of expert readers’ judgments of errors types (e.g., Beason, 2001), or arguments for viewing errors as useful indicators of students’ learning and development as writers (e.g., David Bartholomae’s The Study of Error and Barry Kroll and John Schafer’s Error Analysis).

Compositionists’ calls for evidence-based, replicable research in support of teaching practices in writing classes not only reveal the lack of empirical studies in composition, but indirectly highlight the continued marginalization of research taking place within the larger discipline that includes multilingual students and researchers. Writing instructors not familiar with scholarship in second-language writing studies might be surprised to learn of the ongoing and contentious debate regarding the effectiveness (or not) of providing feedback on students’ grammar in their writing assignments. Since assumptions about the effectiveness, or ineffectiveness, of corrective feedback are relevant to all writing classes (including WAC courses), and since composition instructors are increasingly likely to count multilingual learners among their students (especially in two-year college settings), it is helpful for instructors to be familiar with findings from this expansive body of research. Many of the published papers, which include empirical studies, reviews, and meta-analyses, respond, directly or indirectly, to John Truscott’s 1996 article, The Case against Grammar Correction in L2 Writing Classes. Drawing on a comprehensive and detailed review of studies from the related fields of composition, second-language writing, and applied linguistics, Truscott vigorously discredits the effectiveness of providing corrective feedback as a means to improving students’ writing, reiterating Braddock et al.’s conclusion that such feedback is not only ineffective but harmful to students and their development as writers. According to Truscott, “no valid reasons [exist] for continuing the practice” (360). Embedded within Truscott’s argument is a call for researchers to examine critically the practice of providing corrective feedback, as well as its potential side effects, rather than accept without question its effectiveness.

Truscott’s recommendation to immediately abandon corrective feedback in second-language writing classes, along with his call for research-supported practices, has currently generated over 300 papers published on this topic alone, making it one of the most researched areas in second-language writing studies. One of the first and most prolific researchers to respond to Truscott has been Dana Ferris, whose research offers counter-evidence to Truscott’s claims through examinations of different types of corrective feedback on a wide range of linguistic forms and structures (The Case For Grammar Correction, The ‘Grammar Correction’ Debate, and Does Error Feedback Help Student Learners?). Contrary to Truscott’s claims, the vast majority of studies conclude that corrective feedback is beneficial, at least for multilingual learners. Truscott maintains, however, that conclusions in favor of corrective feedback are misleading or inaccurate. Despite heated disputes regarding the interpretation of the findings, Ferris and Truscott agree that research to date on the topic is insufficient, and that the burden of proof lies with proponents of the practice.

Interpreting the research on corrective feedback is challenging in part due to the wide range of feedback practices studied (see Ferris’ Written Corrective Feedback, for an excellent review and annotated bibliography). For example, feedback can be direct or indirect, depending on if the correct form is provided (direct) or if a mistake is simply marked as such without the correction (indirect). Feedback can also be focused or unfocused, with focused feedback limited to a few grammatical structures (usually those relevant to recent instruction or genre demands) and unfocused feedback unlimited in this respect. Some studies compare feedback on form versus feedback on content. Finally, most studies are limited to short-term versus long-term effects. In general, the findings tend to favor focused, indirect feedback for positive effects on students’ writing. As scholars have noted, however, the number of potential variables in these studies greatly limits the strength of their findings, as it is difficult to determine which variable(s) contributed to the (positive or negative) effects in any given study.

Recently, researchers from applied linguistics have entered the debate with an expanding body of meticulously designed quasi-experimental studies on the effectiveness of corrective feedback. To minimize limitations related to the number of variables, and maximize the strength of their findings, researchers in this current strand intentionally limit the scope and focus of feedback to a single linguistic category or set of structures. One particular set of grammatical structures that has been the focus of many studies includes the English articles a/an and the. While this choice may seem odd given the complexity of the English article system, in many respects a focus on articles is ideal as articles are unavoidable in English, difficult for learners across proficiency levels, and rule-governed in many cases. Moreover, Ferris classifies students’ errors with articles as “treatable,” meaning they can be simplified enough to be teachable (Treatment of Error). The specific rule these studies are centered around is the rule stating that the indefinite article a/an is used for first-mention nouns (the first time a singular count noun is mentioned in a text) and the definite article the is used for subsequent mentions of the same noun. (I saw a cat climb a tree. The cat seemed to get stuck in the tree,) Thus, the argument for examining articles in feedback studies is that articles are unavoidable in certain genres in English, and are observable, teachable, and measurable in students’ writing; very few grammatical structures meet all of these requirements.

In addition to limiting the variable of grammatical focus, studies in this quasi-experimental tradition are also designed to include one or more specific treatment groups and a control group. The treatment group includes direct corrective feedback or direct corrective feedback with some type of metalinguistic explanation (oral, written, or both), and the control group receives no corrective feedback or feedback unrelated to grammatical accuracy. Results from these studies from applied linguistics show a positive effect for the treatment groups as compared to the control groups in post-tests of grammatical accuracy, and thus have been interpreted in support of corrective feedback, at least for the structures examined and the types of students included.

Many interpret findings from these latest studies as counter-evidence to Truscott’s claims that corrective feedback has a minimal or negative effect on students’ accuracy. A close look at the specifics of the studies countering Truscott, however, reveals problems and raises doubts as to the strength of the conclusions that can be drawn from them. Noting that there are exceptions to the article rules included in their study, John Bitchener and Ute Knoch decided not to provide corrective feedback in cases not fitting the first mention-subsequent mention rule, stating that “because there are occasions when the definite article is required for referring to something for the first time ... or for referring to mass nouns, [corrective feedback] was not provided on such occasions” (202). That is, the feedback provided to the experimental groups was intentionally incomplete, leaving the question open as to the true effects of incomplete feedback focused on just one aspect of the grammatical structure in question. Moreover, Ellis and colleagues hypothesized that “one effect of the correction might have been to signal to learners that they needed to use articles a lot and may have led to errors of overuse. ... It is possible that the correction led to overuse of articles in contexts that were not the focus of this study and that did not require the use of an article but this remains an issue for further study” (369). Once again, by not accounting for all articles used by their participants the researchers left the question open as to the true effects of one-rule focused grammar feedback.

In sum, while several recent studies have shown participants’ writing to improve in accuracy after receiving some type of corrective feedback on grammar, to our knowledge, none of the existing studies takes into account potential negative effects on participants’ accuracy, even when these effects are limited to the form under investigation. To date, therefore, researchers still have not fully addressed Truscott’s concern that corrective feedback on students’ grammar may be not only useless, but actually harmful if potential side effects are not counted along with other effects. The current study aims to address the issue of potential side effects that were identified, but underreported, in previous research. It does so by replicating the feedback methods used in studies frequently cited in favor of corrective feedback. Specifically, following studies by Bitchener and Knoch, Ellis, et al., and Younghee Sheen, we examined the effects of providing a teachable subset of rules on article usage on participants’ accuracy in their subsequent writing. In doing so, the study contributes to the ongoing debate as to the effectiveness of corrective feedback for writing pedagogy. Additionally, the study responds to Haswell’s call for research in composition that is replicable, aggregable, and data-supported.

The Current Study

Responding to calls for evidence-based research and replications of earlier studies to support or question practices in composition courses, as well as our own concerns about the strength of findings in support of corrective feedback, our study addresses the following research questions:

1. Does corrective feedback provided on a specific grammatical structure lead students to produce fewer errors of that structure in subsequent writing tasks?

2. What is the difference, if any, between feedback with and without metalinguistic explanation on students’ grammatical accuracy in subsequent writing tasks?

3. Does corrective feedback provided lead to any negative effects on students’ grammatical accuracy in subsequent writing tasks?

Answers to these questions have the potential to greatly affect both students and instructors, as students attempt to develop greater linguistic control in their writing, and as instructors consider what type of response to students’ writing will lead to positive effects. If providing feedback on grammar is not effective in improving students’ writing, then many writing instructors may need to reconsider how much and what type of feedback they are providing. If, on the other hand, this feedback is effective, then what are its effects?

Method

Participants

To provide a fair comparison with existing studies, participants for the current study were also multilingual students. A total of 63 students from three sections of writing classes at a large, urban, two-year college participated in the study, among them 18 graduates of US high schools and 45 from overseas high schools (see Kristen di Gennaro’s Heterogeneous and Assessment of Generation 1.5 for more about the heterogeneous student population in composition courses). There were 31 male and 32 female students aged 18-45, taking basic writing with the same instructor. As the study involved human participants, we requested and were granted an expedited review under Category 7 from the college’s institutional review board to conduct this study (IRB File #2015-0663).

Grammatical Structures and Study Instruments

Following previous research on corrective feedback, we examined students’ use of a/an and the in their writing. Specifically, for treatment we focused on use of the indefinite article a/an preceding first-mention nouns and the definite article the preceding subsequent mentions of nouns as in Jane bought a ring and a necklace for her mother’s birthday. Her mother liked the ring, but hated the necklace.

To minimize potential differences in students’ responses based on genre expectations, and to maximize occasions for students to use the grammatical structures that were the focus of feedback, we selected two types of narrative-based tasks, a writing task and a correction task. We adopted the procedures used by Ellis and colleagues to elicit students’ writing for the writing task. Specifically, participants were given a set of cartoon-based illustrations with an accompanying written narrative for students to read. The written part of the task was then collected and students were instructed to re-tell the story with as much detail as possible, using the illustrations as a guide. The cartoons involved multiple individuals and objects introduced and re-introduced in the stories, thus creating ideal contexts for frequent application of definite and indefinite articles. Participants had 30 minutes to complete this task. During the treatment stage of the study, prior to completing the writing task, participants received feedback on the narratives they had composed and were given time to examine their individual feedback. The type of feedback participants received was based on the group they were assigned to (see Table 1 for feedback types).

A narrative-based genre was also ideal for the correction task as both definite and indefinite articles appear frequently in narratives. For this task, participants were given a written fable, with illustrations, to read. Participants were instructed to insert missing words wherever they thought words were needed, with no blanks provided in the text. An example with short but essential function words (i.e., prepositions, pronouns) but not articles was provided (see Ekiert and di Gennaro for further details about the instruments in the study).

Procedures of the Study

For the sake of comparability, the current study replicated key features from previous studies. Specifically, we adopted a quasi-experimental design in which students from existing classes were divided into three groups. As shown in Table 1, Group 1 (n=22) received direct corrective feedback on a specific grammatical structure, Group 2 (n=23) received direct feedback and metalinguistic explanation on the same grammatical structure, and Group 3 (n=18) served as a control group in which students received general feedback in the form of end notes on their writing and nothing specific to the grammatical structure that was the focus of the feedback given to the other two groups.

Table 1. Groups and Treatment Types

Group	Treatment Type	Example of Treatment
Group 1 (n=22)	Direct error correction on grammatical structure	Incorrect uses with a and the were corrected above each error; a or the were inserted where missing (only first mention a and subsequent mention the were corrected).
Group 2 (n=23)	Direct error correction and metalinguistic explanation on grammatical structure	Incorrect uses with a and the were corrected above each error; a or the were inserted where missing (only first mention a and subsequent mention the were corrected). In addition, students received the following explanation attached to their piece of writing: Use “a” when referring to something for the first time. Use “the” when referring to something that has already been mentioned. Illustration of the rule taken from the writing task: A man and a woman went to a restaurant for dinner. The man ordered a bottle of wine and the woman drank the wine.
Group 3 (n=18)	End notes; No corrective feedback on grammatical structure	Students received summary end notes on the overall quality of their writing. Nice narration! Very colorful and full of detail. There is even a lovely ‘lesson to be learned’ at the end of your short narration. You have captured enough details for a reader to appreciate the humorous aspects of this story. In the future, you may add a lesson to be learned from a story like that (the point) at the end of it, not the beginning. Be careful calling it “a novel”—it was just a picture story. No in-text corrections were provided. No reference to article uses were made.

Students from all three groups provided writing samples from a pre-test, immediate post-test, and delayed post-test over the course of one semester. As summarized in Table 2, the pre-test was administered in week 1, three treatment sessions took place in weeks 3, 4 and 5 with an immediate post-test also in week 5, and a delayed post-test was administered in week 11.

Table 2. Study Design and Schedule

Week	Stage in the Study	Activity
1	Pre-Test	Writing Task 1 Correction Task 1
3	Treatment	Feedback on Writing Task 1 Writing Task 2
4	Treatment	Feedback on Writing Task 2 Writing Task 3
5	Treatment Immediate Post-Test	Feedback on Writing Task 3 Writing Task 4 Correction Task 2
11	Delayed Post-Test	Writing Task 5 Correction Task 3

Previous analyses of the effects of corrective feedback on students’ use of articles have focused on the selective use of a/an or the where a/an is required for first-mention of nouns and the for subsequent mentions, as this is the focus of the corrective feedback provided in the studies. Students’ use of articles in other contexts was intentionally ignored and not counted by previous studies. By our estimates, this narrow focus on just one rule for the use of a/an and the overlooked a third of all articles used by the students in the writing task. Thus, accounting for all articles used by the writers in the study was crucial if we were to provide a more complete measure of participants’ accuracy and the effectiveness of feedback on grammar. In fact, in reviewing these previous studies, we wondered if providing students with this teachable, yet over-simplified, feedback about article usage might inadvertently lead students to produce more, not fewer, article errors in their writing. Thus, in addition to counting participants’ accurate use of articles where the feedback was provided, we also counted their use of articles where the feedback was not provided to see if their accuracy with articles may have been negatively affected by the intentionally incomplete corrective feedback they had received.

To calculate participants’ accuracy on writing and correction tasks, we counted the number of cases where participants accurately provided the correct article where required and divided this by the total number of required cases. We converted this number to a proportion of 100. Similarly, to calculate article overuse we counted the total number of overused articles divided by the total number of required cases and expressed it as a proportion of 100.

Results

The results of our analyses are presented in Figure 1. The figure includes mean accuracy scores for all three groups (direct feedback, direct feedback and metalinguistic explanation, and end note group) at three points in time: for the pre-test, immediate post-test, and delayed post-test.

Figure 1 shows how each group performed with regard to their accuracy on all uses of articles a/an and the; that is, mean scores in Figure 1 include participants’ accuracy on the articles that were the focus of correction as well as articles that were not the focus of correction. As depicted in Figure 1, all three groups started with very similar scores at the time of the pre-test and all three groups ended with scores similar to one another by the time of the delayed post-test. Notably, the group that received additional metalinguistic explanation performed worse than the other two groups, hardly improving at all at the time of the immediate post-test. By the time of the delayed post-test, however, this group improved more, and performed similarly to the other feedback group, illustrating similar improvement over time for accuracy on all article uses, those that were and were not the focus of the study. Perhaps surprisingly, the group that received no instruction on articles, only end notes on the writing task, outperformed both corrective feedback groups.

In Figures 2 and 3 that follow, we assess, separately, participants’ accuracy on articles in two categories—articles for which the writers received feedback (i.e. rule-based first and subsequent mention of a/an and the) and the remaining articles (e.g., the types of uses that require the for referring to something for the first time or for referring to mass nouns). Figure 2 shows how each group performed with regard to the grammatical forms that were the focus of instruction, first-mention a/an and subsequent mention the, and thus, provides a point of comparison with previous studies focused on these same forms and functions. As seen in Figure 2, on average, all groups performed similarly on the pre-test, confirming their comparability at the start of the study. All groups also improved in accuracy for the immediate post-test, with the group that received additional metalinguistic explanation improving the least of the three groups. By the time of the delayed post-test, however, all three groups were again close together in terms of accuracy for the specific grammatical forms that were the focus of instruction. These findings are not surprising, as they agree with those found in previous studies.

Figure 3 shows how each group performed only with regard to their accuracy on additional uses of articles a/an and the, that is, article uses that were not the focus of instruction provided in the study. Again, pre-test scores indicate that all three groups were comparable at the start of the study. Worth noting in Figure 3 is that the two feedback groups declined in accuracy at the time of the immediate post-test, with the group receiving the most feedback performing the worst of the three groups. By the time of the delayed post-test, this group had recovered, displaying improvement overall, but the other feedback group declined even further, displaying less accuracy over the course of the study. Figure 3 indicates that participants did not improve as noticeably with grammatical forms that were not the focus of instruction. It also suggests that feedback on one rule for the use of articles may have led to decreased accuracy on non-rule-based uses of articles, perhaps as a result of the feedback. Once again, the end note group, who did not receive feedback on the specific grammatical forms and uses that were the focus of the feedback, performed the same as or better than the two feedback groups.

For the sake of completeness, we also calculated overuse patterns across the three groups, that is, cases in which participants used articles where they were not required, including contexts that were not the focus of the feedback. As shown in Table 3, while overuse of articles occurred across all three groups, tellingly, the group that received the most feedback (both direct correction on their writing as well as rule explanation with examples) recorded steady increases of article overuse (from 0.04 to 0.06), again raising doubts as to the helpfulness of feedback that may be detailed and clear but also incomplete with regard to how a linguistic system works.

Table 3. Overuse of articles a/an and the

Group	Pre-test (%)	Immediate Post-test (%)	Delayed Post-test (%)
DF	0.05	0.04	0.05
DF+ME	0.04	0.05	0.06
EN	0.03	0.04	0.04

Note. DF=direct feedback; DF+ME=direct feedback and written metalinguistic explanation; EN=end notes

Discussion

The results of our analyses show that feedback focusing on rule-based, teachable uses of the articles a/an and the had a negative impact on participants’ accuracy with other, less teachable, uses of these articles. Furthermore, the end note group serving as a control group outperformed or matched the two feedback groups in terms of accuracy on all uses of articles, those that were and that were not the focus of feedback.

In response to our first research question, Did corrective feedback provided on a specific grammatical structure lead students to produce fewer errors of that structure in subsequent writing tasks?, we found that yes, students’ accuracy for the specific grammatical structures that were the focus of feedback improved over the course of the study, albeit moderately. This was the case for all three groups of students. This finding agrees with the vast body of research countering claims that feedback can lead to negative or harmful effects. In terms of the type of feedback provided, in answer to our second research question, What was the difference, if any, between feedback with and without metalinguistic explanation on students’ grammatical accuracy in subsequent writing tasks?, we found that the group of students who received additional metalinguistic explanation did not perform consistently better than the other two groups who did not receive this enhanced feedback, suggesting that more is not always better when it comes to explaining grammar for improving students’ accuracy in writing. Finally, in response to our third research question, Did the corrective feedback provided lead to any negative effects on students’ grammatical accuracy in subsequent writing tasks?, we found that, yes, the two groups of students who received feedback initially declined in accuracy on article uses that were not the focus of the study, with one group’s accuracy declining even further for the delayed post-test. We also observed a systematic and growing trend of overuse exhibited by the group of students who received additional metalinguistic explanation. This finding suggests that claims that some feedback may do more harm than good is not without warrant.

The current study calls attention to the danger in supporting or dismissing feedback without attention to the type of feedback and potential effects on students’ developing understanding of writing and written language. While corrective feedback had a positive effect on students’ accuracy for the grammatical forms and uses that were the focus of instruction, the same corrective feedback had a negative effect on students’ accuracy for other uses of the same forms. Is increased accuracy for some uses worth the potential accompanying decreased accuracy for other uses? Future research in composition might consider focusing on finding the optimal balance of feedback to provide on students’ writing, how much and of what type, rather than dismissing outright a focus on grammar.

Another striking finding from the current study is that all groups, including the end notes group serving as a control group, improved in accuracy on the specific grammatical form and function that were the focus of feedback, even though the end note group did not receive feedback on this form and function, only general feedback on their writing. In fact, the end note group performed the same or better than both groups receiving corrections in the delayed post-tests for all structures and uses studied. Researchers and instructors who remain skeptical that corrective feedback is harmful may consider instead that other factors, besides corrective feedback, may lead to improved accuracy in students’ writing over time. For example, it is possible that having students engage in writing tasks in which certain grammatical structures arise naturally and frequently may be both necessary and sufficient to improve students’ accuracy in using those structures. That is, the nature of the tasks students are asked to write, and the accompanying cognitive processes students engage in while completing these tasks, may be more influential on students’ accuracy than the nature of the feedback teachers provide on students’ final written products. Future research on students’ accuracy in writing might broaden the scope of study to look beyond feedback on individual products to the larger field of the types of reading and writing tasks students engage in on a regular basis.

As with all studies, the current study comes with a set of limitations. The focus on only one grammatical construction limits the generalizability of our findings to other areas of grammar. A future line of research in composition might keep the study design the same but select additional grammatical constructions in order to determine which errors should be the target of grammatical correction in composition classrooms. Another limitation of the present study to be addressed by future research could focus on the nature of metalinguistic information given to students in composition classes, considering how strong the effects of such feedback are. Finally, the exclusive focus on multilingual participants in our study only partially reflects the demographic make-up of WAC and composition courses that the readers of Composition Forum teach. Thus, the present findings should be interpreted within the context of the scope of the study.

Implications and Conclusion

Several implications can be drawn from the current study. In terms of writing instruction, our results suggest that composition instructors, and others who teach writing, consider how realistic their expectations are with regard to students’ uptake of instructors’ feedback in their subsequent writing. Specifically, since feedback that appears useful in some instances may be less useful in other, unexpected areas, composition instructors might take into account potential, unintended consequences of their feedback, especially feedback on grammar. Feedback that is detailed and clear is likely still incomplete with regard to the larger linguistic system, making students’ interpretations of feedback unpredictable. Moreover, structures that are teachable might not be equally learnable.

Another implication related to grammar instruction follows from the control, end note group’s performance, which equaled or surpassed that of the instructional groups, suggesting that the nature of the writing task and the summative feedback were sufficient to lead to improvement in participants’ writing. When designing and selecting writing assignments, instructors might want to spend some time analyzing the linguistic features typically inherent in the tasks they select, and adjust their expectations or instruction accordingly. For example, the narrative task adopted in the current study included natural occurrences of the linguistic structure that was the focus of instruction. Rather than dedicate time to providing detailed feedback on grammar, instructors may spend their time more productively selecting tasks likely to elicit the linguistic forms they want students to use. In a similar vein, when choosing the form of response to student writing, teachers may suppress the urge to provide in text corrections and, instead, offer summative end notes. A possible explanation for the end note group’s superior performance in our study may reside in the fact that while end notes included both positive and negative comments (see Table 1 for examples), corrective feedback on student writing was only negative. Thus, it is quite possible that the mix of positive and negative feedback in the end notes had some impact on the results.

The current study also leads to methodological implications in composition research. While we did not set out to challenge or support strong claims that feedback on students’ writing is not only useless but harmful, our results force us to acknowledge that these claims have not been fully addressed, despite repeated claims to the contrary. Indeed, the results of our study indicate that the effectiveness of feedback depends, in part, on how accuracy is defined. Studies showing positive effects on accuracy rely on a limited definition of accuracy. When accuracy is defined as correctness in all cases, and not just in those cases selected by the researcher as relevant, results from previous studies are less convincing. This is not to say that we promote a complete ban on providing students feedback on grammatical forms and functions. Rather, we hope that the results of our study motivate compositionists to continue to seek effective feedback methods but with greater attention to unexpected consequences on students’ developing understanding of writing and written language.

An additional implication concerns the value placed on replication studies and the marginalization of research on multilingual writers in composition. We hope that compositionists—those who believe corrective feedback to be harmful, those who find it beneficial, and those who are on the fence—will respond to Haswell’s call for replicable, aggregable, and data-supported research and contribute to the ongoing conversations that, until now, have been marginalized in studies of multilingual students even though they are potentially applicable to all students in composition courses. Thus, this study addresses the efficacy of grammar instruction as related to feedback given to multilingual student writers with an eye toward implications for students whose primary language is English. While the current study addresses descriptive grammar forms, that is, linguistic forms that most native-English speakers would find intuitive and unproblematic, similar results may hold for the type of prescriptive grammar forms often addressed to student writers of all backgrounds, such as rules about punctuation and other writing conventions. Results of the study also support Inoue’s and Young’s recent critiques of traditional approaches to evaluating—and valuing—grammatical accuracy in composition courses.

Finally, we would like to leave readers with the encouraging discovery that students who did not receive detailed feedback on grammatical forms, but who engaged in writing tasks and genres in which certain forms are prevalent, improved in accuracy using those forms. In other words, instead of providing detailed feedback on students’ grammatical accuracy in writing, instructors’ time and effort may be better spent on considerations of task design and genre selection as these may be more influential in developing students’ writing than feedback on grammar.

Acknowledgements: The authors would like to thank K. Philip Choong for his assistance with instrument design. We are also indebted to the editors of Composition Forum and the anonymous reviewers who provided helpful suggestions on the manuscript. All remaining mistakes are ours. This research was supported in part by PSC-CUNY Research Award TRADA-46-174.

Works Cited

Allan, Elizabeth G. and Dana Lynn Driscoll, D. R. Hammontree, Marshal Kitchens, and Lori Ostergaard. The Source of Our Ethos: Using Evidence-Based Practices to Affect a Program-Wide Shift from ‘I Think’ to ‘We Know.’ Composition Forum, vol. 32, Fall 2015, n.p. https://compositionforum.com/issue/32/oakland.php.

Anson, Chris M. The Intelligent Design of Writing Programs: Reliance on Belief or a Future of Evidence. WPA: Writing Program Administration, vol. 32, no. 1-2, 2008, pp. 11-36.

Bartholomae, David. The Study of Error. College Composition and Communication, vol. 31, no. 3, 1980, pp. 253-69.

Beason, Larry. Ethos and Error: How Business People React to Errors. College Composition and Communication, vol. 53, no. 1, 2001, pp. 33-64.

Bitchener, John and Ute Knoch (2010). The Contribution of Written Corrective Feedback to Language Development: A Ten Month Investigation. Applied Linguistics, vol. 31, 2010, pp. 193-214.

Blaauw-Hara, Mark. Why Our Students Need Instruction in Grammar, and How We Should Go about It. TETYC, vol, 34, no. 2, 2006, pp. 165-78.

Braddock, Richard, Richard Lloyd-Jones, and Lowell Schoer. Research in Written Composition. Champaign, IL: NCTE, 1963.

Cole, Daniel. What if the Earth Is Flat? Working With, Not Against, Faculty Concerns about Grammar in Student Writing. The WAC Journal: Writing Across the Curriculum, vol. 25, 2014, pp. 7-35.

di Gennaro, Kristen. Assessment of Generation 1.5 Learners for Placement into College Writing Courses. Journal of Basic Writing, vol. 27, 2008, pp. 61-79.

di Gennaro, Kristen. The Heterogeneous Second-Language Population in US Colleges and the Impact on Writing Program Design. Teaching English in the Two-Year College, vol. 40, 2012, pp. 57-67.

Ekiert, Monika, and di Gennaro, Kristen. Focused written corrective feedback and linguistic target mastery: Conceptual replication of Bitchener and Knoch (2010). Language Teaching, 2019, pp. 1-19 https://doi.org/10.1017/S0261444819000120.

Ellis, R., Y. Sheen, M. Murakami & H. Takashima. The effects of focused and unfocused written corrective feedback in an English as a foreign language context. System, vol. 36, 2008, pp. 353-371.

Ferris, Dana R. The Case for Grammar Correction in L2 Writing Classes: A Response to Truscott (1996). Journal of Second Language Writing, vol. 8, 1999, pp. 1-10.

Ferris, Dana R. Treatment of Error in Second Language Student Writing. U of Michigan Press, 2002.

Ferris, Dana R. The ‘Grammar Correction’ Debate in L2 Writing: Where Are We, and Where Do We Go From Here? (And What Do We Do in the Meantime...?). Journal of Second Language Writing, vol. 13, 2004, pp. 49-62.

Ferris, Dana R. Does Error Feedback Help Student Learners? New Evidence on the Short- and Long-Term Effects of Written Error Correction. In K. Hyland & F. Hyland (Eds.), Feedback in second language writing: Contexts and issues (pp. 81-104). Cambridge UP, 2006.

Ferris, Dana R. Written Corrective Feedback in Second Language Acquisition and Writing Studies. Language Teaching, vol. 45, 2012, pp. 446-59.

Hartwell, Patrick. Grammar, Grammars, and the Teaching of Grammar. College English, vol. 47, no. 2, 1985, pp. 105-27.

Haswell, Richard H. NCTE/CCCC’s Recent War on Scholarship. Written Communication, vol. 22, no. 2, 2005, pp. 198-223.

Haswell, Richard H. Minimal Marking. College English, vol. 45, 1983, pp.600-04.

Inoue, Asao. How Do We Language So People Stop Killing Each Other, Or What Do We Do About White Language Supremacy? Conference of College Composition and Communication Plenary Address, March 14, 2019.

Kroll, Barry, and Schafer, John C. Error Analysis and the Teaching of Composition. College Composition and Communication, vol. 29, no, 3, 1978, pp. 242-248.

MacDonald, Susan Peck. The Erasure of Language. College Composition and Communication, vol. 58, no. 4, 2007, pp. 585-625.

Lunsford, Andrea A. and Lunsford, Karen J. ’Mistakes Are a Fact of Life’: A National Comparative Study. College Composition and Communication, vol. 59, no. 4, 2008, pp. 781-806.

Shaughnessy, Mina P. Errors and Expectations: A Guide for the Teacher of Basic Writing. Oxford UP, 1977.

Sheen, Younghee. The Effect of Focused Written Corrective Feedback and Language Aptitude on ESL Learners’ Acquisition of Articles. TESOL Quarterly, vol, 41, 2007, pp. 255-83.

Truscott, John. The Case Against Grammar Correction in L2 Writing Classes. Language Learning, vol. 46, 1996, pp. 327-69.

Young, Vershawn A. Should Writers Use They Own English? Iowa Journal of Cultural Studies, vol. 12, 2010, pp. 110-117.