Providing feedback is viewed both by teachers and students as an important part of writing instruction. One type of feedback that writing teachers provide is error correction. It is perhaps the most widely used method for responding to student writing. How teachers correct second or foreign language students’ writing is a topic that has attracted teachers and researchers alike. Ferris (1999) speculated that error treatment is beneficial for adult learners who learn a second language in a formal situation. However, this contention is strongly criticized by a number of scholars who assert that students’ need for error correction is not necessarily indicative of the effectiveness of such feedback (Truscott, 1999). Ferris (1999), while acknowledging that provision of feedback is a rather time-absorbing and exhausting aspect of teacher’s profession,  averrs  that students do not become more proficient writers just by reading and writing; rather they need some form of feedback to see how others think of their writing.

By the same token, many scholars assert that although a great deal of learning takes place through exposure to comprehensible input, learners may require negative evidence (i.e. information about  grammaticality), in the form of either feedback on error or explicit instruction, when they are unable to discover through exposure alone how their interlanguage differs from the L2 (Bley-Vroman, 1986, cited in Panova & Lyster, 2002 ). “If the corrective feedback is sufficiently salient to enable learners to notice the gap between their interlanguage forms and target language forms, the resulting cognitive comparison may trigger a destabilization and restructuring of the target grammar” (Ellis, 1994, cited in Panova & Lyster, 2002, p. 574). Although teacher feedback seems an ideal and the most preferred one by many students in second language instruction, its fruitfulness in developing students’ writing remains unclear. Several studies, however, directly point to the students’ ignorance, misunderstanding, or disability in applying teacher comments to their subsequent writing (Ferris, 1999). This study was accordingly designed to reveal whether teachers, in providing students with comments to revise their writing, can properly reconstruct the erronous sentences produced by student writers. In other words, the study was meant to discover whether teachers are successful in providing the ‘right’ type of  feedback, the feedback that would help students re-write their own intended meaning, not the meaning imposed by the teacher.

LITERATURE REVIEW

Truscott (1996) contends that grammar correction has no room in a writing course and should be removed from it. He asserts that there is no convincing research evidence that error correction ever helps student writers improve the accuracy of their writing. He argues by reasoning that it does away with second language acquisition (SLA) theories about the gradual and complex process of learning second language linguistic structures and forms.

Ferris (1999, 2004), on the other hand, asserts that teachers, students, or researchers who have criticized corrective feedback (CF) as being ineffective or even harmful may have failed to understand the issue.  It is a part of much larger SLA  process, and like anything else in language acquisition, CF takes time to be effective.

The research evidence on the effects of error correction on students’ writing skills is far from conclusive (Ellis, 2013). Studies (Leki, 1990, for example) investigating the effects of different types of feedback on students’ writing have suggested that explicit correction on surface level errors (spelling, punctuation, grammar) seems to be generally ineffective. Several other studies have demonstrated that provision of feedback on local issues or low level concerns does not assist learners to commit fewer local errors than providing no feedback on such issues does (Truscott, 1996), and some studies have even proposed that correcting local errors leads to making more errors on subsequent drafts (see Truscott, 1996). Truscott (1996) contends that such feedback may not be helpful because students need much longer time to automatize or internalize grammatical rules than would happen from one draft to another.

An overwhelming majority of teachers are still slaving, to use Hariston’s (1986 in Lee, 2009) term, over student writing with little or no avail.  We all witness, even nowadays, that in a great number of teaching contexts “teachers are still slaving over student writing, making deadline after deadline to provide timely feedback to student writing. While teachers burn midnight oil to mark student writing, students who make significant progress as a result of teacher feedback may be few and far between” (Lee, 2009, p. 34; cf. Lee, 2008). Although responding to student writing is an important and meaningful area of teachers' work, it is often described in negative terms, referred to as “frustrating, grueling and anxiety-ridden, tedious and unrewarding. Teachers, despite their efforts, are described in disparaging terms — as composition slaves  and as paternalistic figures who appropriate student writing” (Lee, 2008, p. 1; see Brannon & Knoblauch, 1982; Greenhalgh, 1992; Neal et al., 2007).

More often than not, teachers are very unsure of the consequence of their endeavors, and they are hard pressed to substantiate that students betterment, if any, can be attributed to their diligence and assiduousness (Lee, 2009). The findings of many experimental studies on written corrective feedback conducted over the last 20 years have been so conflicting that second language teachers looking to uphold the instructional choice to correct, or not to correct, the grammar of their students’ written activities are “left in the midst of controversy” (Guenette, 2007, p. 40).

Most of the time, teachers do not understand students’ intentions, or misinterpret their intentions, the result being that teachers, by incorporating their perceived corrections and comments, render, in the first place, the students’ written texts incoherent, and in the second place, make error correction an act of no or little avail. Teachers view themselves as  authorities, intellectually mature, and rhetorically more experienced than their apprentice writers. In classroom writing situations, the reader (i.e. teacher) “assumes primary control of the choices that writers make, feeling perfectly free to correct those choices any time an apprentice deviates from the teacher-reader’s conception of what the developing text ought to look like or ought to be doing. Student writers, then, are put into the awkward position of having to accommodate not only the personal intentions that guide their choice-making, but also the teacher-reader's expectations about how the assignment should be completed” (Brannon & Knoblaunch, 1982, p. 158).

PURPOSE OF THE STUDY

This study aimed to investiagte teachers’ reconstruction of students’ intentions embeded in the erroneous utterances by relying on their intuition or sheer hunches on “empirical footing” (Hamid, 2007, p. 2).  In so doing, it presented erroneous sentences (both context-bound and context-free) to teachers for the purpose of reconstructing them to investigate the role of context in the process of reconstruction.  More specifically, this study sought to find answers to the following questions:

1. Do teachers accurately understand students’ intentions in their erroneous and idiosyncratic utterances?
2. Would context be of any help in reconstructing students’ erroneous sentences correctly?

METHOD

Participants

The participants in this study were nine university English language teachers, of whom one was female and eight males. Their teaching experience ranged from a minimum five years to a maximum thirty two. Of the participants, six held PhD degrees and three MA. They were teaching in the Azad and Payam Noor Universities in the northern part of West Azerbaijan Province, Iran. Their selection was guided by their willingness and convenience to participate.

Moreover, the participants in this study were thirteen English students (nine males and four females). Their selection was based on their availability and willingness to participate. The participating students  were about to complete their BA degrees. They had successfully passed writing related courses such as ‘English grammar 1 & 2’, ‘advanced writing’, ‘essay writing’ and ‘reading courses’. Their ages ranged from  21 to 27.  Their English proficiency level ranged from pre-intermediate to intermediate. On the researchers’ request, they produced writing drafts on the topic of ‘Learning English helps me find a good job’. Guided by the knowledge that the more the number of the participants (i.e. teachers and students), the more reliable the generalizability of the results would be, there was a strong tendency and inclination to include a greater number of teachers and students in the experiment. Upon perceiving their reluctance, notwithstanding the fact that they were in advance assured of the confidentiality and anonymity of the data they produce, we were coerced to be satisfied with those who voluntarily voiced their readiness to cooperate.

Instrumentation

The main source of data in this qualitative–quantitive study came from the students’ write-ups.That is, they willingly volunteered, upon researchers’ decision not to make use of in-class written assignments, to produce new drafts on the topic assigned by the researchers. Since the aim of the study was to measure the extent to which the university English teachers can manage to correctly guess the learners’ intended meanings in faulty (conceptually and  grammatically ambiguous) sentences, with this purpose in mind, the students’ papers were collected and studied with a lot of care and precision with an eye on  the segments of discourse (be it a sentence or a paragraph) which were believed would challenge the teacher-readers’ ability in unearthing the student- writers original intention  masked in the unintelligible piece of  discourse.

The researchers,  having read the students’ write-ups, chose papers containing conceptually  impaired and structurally deformed sentences. Afterwards, they produced a neater and fair copy of students’ drafts  in which for the ease of attention and focus, they applied  a colored pen to highlight thirty nine sentences serving the research purpose (i.e. being semantically and structurally unintelligible).

 It is worth noting that the researchers refused to subject the students’ drafts to  any further modifications or  copy-editing practices in order to keep the naturalness and authenticity of language produced intact. This is what was rightly confirmed by Santos (1988) who asseverated that “artificially prepared passages allow for maximum control of the variables by the researcher, but they also sacrifice the natural quality of unaltered connected discourse” (p. 74; see also Khalil, 1985).

Data Collection Procedure

Shortly after having identified problematic pieces of discourse, we met with the student writers for the elucidation of their meaning embedded in those ill-formed utterances. Not being proficient enough in English to get their message across, they were permitted to reiterate their original meaning in their mother tongue. The researchers, then, produced the L2 version of those L1 sentences. In cases, where there was an atmosphere of uncertainty as to the standard English version of, say, a sentence, or a phrase, the help of a third expert was sought. This procedure of reconstructing erroneous sentences and unmasking the writer’s intention embedded in those idiosyncratic utterances at his or her presence is what Corder (1981) technically calls ‘authoritative reconstruction’ as opposed to plausible reconstruction.  

Afterwards, those purposefully specified idiosyncratic sentences were submitted to the participating university teachers to seek the teachers’ plausible interpretation, correction or reconstruction of these sentences. On perceiving that teachers were reluctant to reconstruct these erroneous sentences, we had to try out alternative ways so as to reduce the load of task for the participating teachers.  

To begin with, we decided to produce an alternative standard English version of the student’s wrong sentence and seek the respondent’s judgment on the rightness or wrongness of  reconstructed sentence which was thought to capture the student intention in that erroneous sentence. Perceiving that in the respondent’s mind there may exist other interpretations than the one proposed by the researchers, we disregarded this data collection procedure.  

The idea of presenting the respondents with two or more alternatives of the student’s erroneous sentence was another option we tested. Guided by the knowledge that different respondents may interpret differently a given idiosyncratic sentence, and backed by Bartholomae’s (1980) reasoning that “for any idiosyncratic sentence, however, there are often a variety of possible reconstructions, depending on the reader’s sense of the larger meaning of which this individual sentence is only a part, but also depending upon the the reader’s ability to predict how this writer puts sentences together” (p. 265),this option was sidestepped as well.

The final decision was that the students’ write-ups in which the erroneous sentences had been highlighted and numbered had to be sent out to the teachers for the interpretation or reconstruction of the students’ intention masked in those faulty utterances.

To sum up, in the first phase, students’ papers containing thirty nine context-bound erroneous sentences were given to the nine participating university teachers for the reconstruction of the ill-formed sentences. After a lapse of ten days, we succeeded to  collect the papers containing teachers’s reconstrcuctions.        

The second phase of data collection procedure involved, again, giving those idiosyncratic sentences to the same six university teachers (but this time the erroneous sentences were context-free) to investigate the effect of context on the teachers’ ability to uncover students’ intention embedded in erroneous utterances. It is worth mentioning that in the interval between two phases three teachers refused to cooperate, and were thus put side form the rest of the study. It is also worth noting that submitting context-free erroneous sentences to the participating teachers was late enough (at least one month or so) to make sure of complete erosion of the effect of context from the teachers’ minds.

Besides, attached to the students’ write-ups was a questionnaire containing both demographic information about the student writers such as their native language (L1), the universities they attended, their proficiency level, the courses they had taken and so on, in order to investigate whether this general information would be of any help for teachers in uncovering the students’ intentions. Two general questions aimed at gathering further information so as to deepen and sharpen our understanding of teachers’ ability in text interpretation, reconstruction and error detection were also asked. The questions were as follows:

1. Do you think context would be of help in reconstructing and interpreting the studennts’ intention from the deviant  utterances?
2. Which type of sentence is more difficult to reconstruct: a semantically wrong sentence or a grammatically wrong sentence?

Juxtaposing teachers’ plausible reconstructions with authoritative reconstructions, we can pass judgment on the teachers’ ability or inability in unearthing the students’ intentions in the erroneous sentences.

Data Analysis

The bulk of data for this study came from students write-ups and university teachers’ interpretation of erroneous sentences produced by the students. Since there were a great number of interpretations, we needed a categorization model to classify them. This categorization system was devised copying Khalil’s (1985; see also Bartholomae, 1980) model.   

The procedure is by utilizing a classification system to map teachers’ plausible reconstructions against authoritative reconstructions to judge whether teachers can correctly guess students’ meaning from unintelligible utterances. This categorization system allows the researcher to place the teachers’ reconstructed sentences along the continuum: at one extreme is the option of ‘Not Captured’ and  at the far off extreme is the option of ‘Totally Captured’, and in between fall the options of ‘Minimally Captured’ and ‘Considerably Captured’. For brevity, the four scales are abbreviated and diagrammatically shown along the continuum as follows: Not Captured (NC) Minimally Captured (MC) Considerably Captured (CC) Totally Captured (TC).

It is worth mentioning that as with other categorization systems, one cannot with one hundred percent of certainty assign the reconstructed sentence to appropriate scales; in particular, whether the reconstructed sentence should be subsumed under MC or CC seems to be a personal decision. That is, the element of subjectivity is vivid in this categorization. To moderately overcome this problem, the help of a proficient university teacher was sought.

RESULTS

Research Question One

In this section, data is provided to answer the first research question. The table below presents the students’ original sentences, teachers’ reconstructed sentences (i.e. plausible reconstructions) and students’ intended meanings (i.e. authoritative reconstructions) to pass judgments on teachers’ ability in reading their students’ minds from the idiosyncratic sentences.

Table 1: Students’ original sentences, authoritative reconstructions, and plausible reconstructions

̽ Note: For each erroneous sentence there are at least nine, with a  maximum of, fifteen, plausible interpretations. Owing to space constraints, one plausible interpretation from the teachers is given as an example in the ‘plausible section’ of  Table 1. 

Table 2 provides details of the teachers’ reconstruction with respect to the four categories of Not Captured (NC),  Minimally Captured (MC), Considerably Captured (CC), and Totally Captured (TC).

Table 2: Teachers’ reconstruction of context-bound sentences

Prior to analyzing teachers’ reconstructions, a few points need to be elucidated about the above table. This table shows nine teachers’ attempts at reconstructing contextualized erroneous sentences. So, the numbers opposite each sentence represent the number of teachers. If care is taken, we will notice that the number of teachers who attempted each sentence is the same (that is 9). It does not, however, mean that all teachers attempted all sentences. In fact, some of the sentences were partially reconstructed (i.e. fifteen) and some were totally left unchallenged (i.e. nineteen). To rest assured, whether the partially reconstructed or totally unreconstructed sentences slipped teachers’ attention or they proved to be difficult to interpret, the researchers sought the respondents’ viewpoints as to the unchallenged sentences. They were unanimous in saying that partially reconstructed or unreconstructed sentences proved to be hard to construe. Accordingly, all unreconstructed sentences were subsumed under the NC category, but the partially reconstructed sentences were placed under either NC or MC depending on the degree to which they embraced the student-writers’ intended meaning in the erroneous sentences.

In this study, teachers, on the whole, attempted three hundred and fifty one sentences out of which one hundred and fifteen (that is about 32.76%) displayed teachers’ perfect inability in reading the students' minds. Ninety one sentences (25.92 %) only slightly scratched the students’ intentions. Eighty four sentences (that is about 23.93 %) only thinly missed the target. Sixty one sentences (17.38 %) out of the total number of sentences displayed teachers’ ability in perfectly covering the students’ meaning from the unintelligible sentences. In brief, teachers were unsuccessful in guessing correctly the students’ intention in two hundred and six sentences (58.68 %) and only proved to gain success in one hundred and forty five sentences (41.31%).

There seems that three hundred and fifty one interpretations of the thirty nine sentences were evenly distributed under four categories (i.e, NC,  MC, CC, TC);  the only exceptions are sentences 5, 7, 9, 23, 26, 29 and 39. That is, the interpretations and reconstructions of these sentences are unevenly accumulated under the categories of NC and MC. To put it differently, these sentences proved to be difficult for the teachers to reconstruct. In contrast, the reconstruction of sentences 2, 11, 24, and 36 are unevenly accumulated under the categories CC and TC. That is, these sentences proved to be easy for the teachers to reconstruct.

Research Question Two

The role of context in the teachers’ reconstruction of deviant sentences (research question 2) is dealt with in the following section. With the above analysis of teachers’ reconstructions of context-bound sentences, we move to the analysis of teachers’ ability in reconstructing context-free sentences.

Table 3: Teachers’ reconstruction of context-free sentences

Table 3 briefs reconstruction of the same number of context free sentences (i.e. 39 sentences) by six teachers. In this phase of the study, on the whole, two hundred and thirty four reconstructions  were assigned to the categories NC, MC, CC and TC. As with the reconstruction of context bound sentences, NC category included the most number of sentences. In plain terms, seventy seven reconstructions (32. 90 %) out of two hundred and thirty four belonged to the NC category.  It means that about 32.90 percent of teachers’ reconstructions of erroneous sentences were wide of the mark. Sixty four reconstructions (27.35 % ) belonging to the MC category were also wide of the mark. Putting the reconstructions belonging to the categories NC and MC together, we found that one hundred and forty one reconstructions (60. 25 %) widely escaped the students’ intended meaning. Just ninety three reconstructions (39.74 %, adding up the reconstructions belonging to the categories CC and TC together) captured students’ intended meanings in those sentences. The sentences that proved to be hard for the teachers to reconstruct were items 5, 6, 7, 10, 13, 32, 37 and 39 and only item number 2 was easy to reconsruct.

 Comparison of context-bound and context-free reconstruction tables did not reveal considerable differences. In the context-bound table, about 58.68 percent of the reconstructions widely missed the mark;  this is while in the context-free table, 60.25 percent of the reconstructions were wide of the mark. The reconstructions capturing the students’ intentions were 41.31 percent and 39.74 percent for the context-bound and context-free sentences, respectively. Therefore, there exists a very negligible difference between the context-bound and context-free reconstructions.

The results obtained in this phase of this study not only undervalued the role of context but also contradicted the participating teachers’ expectations.  All of the teachers with respect to the role of the context categorically asseverated that context is very crucial in discovering the students’ intentions but it did not turn out as expected. But the results turned out to be otherwise.

To illustrate the contradiction, examples of teachers’ responses to the role of context are cited as an evidence.

T1: The immediate sentences that precede or follow a troublesome sentence may contain clues that help the reader to decode the intention enclosed in that troublesome sentence or piece of discourse.

T2: The whole context of the discourse may provide hints as to the interpretation of that difficult sentence.

T3: It is utter naivity always to seek the key for the solution of a problematic sentence in the immediate context. Sometimes the key to the problem lies in the far-off context.

To end the analysis of context-bound and context-free reconstructions, the researchers found that teachers did not achieve much success in uncovering students’ intention contained in the idiosyncratic sentences.  Also, the context in this study owing to reasons that will be put forward in the ‘discussion section’ rendered no help to the teachers in guessing the students’ intended meaning from the erroneous sentences.

Getting through the analysis of reconstruction of context-bound and context-free sentences, it is time to deal with the analysis of grammatically and semantically deviant sentences.

Table 4: Teachers’ reconstruction of grammatically and semantically deviant sentences

           Total                                           49.7 %                                               39.14%

As stated in the data collection section, one of the two questions which was posed to the teachers was “Which type of sentence is more difficult to reconstruct, a grammatically deviant sentence or a semantically deviant sentence?”

To this end, all of the thirty nine sentences, in close collaboration  with an expert, with great care and precision were assigned to the ‘Grammatically Deviant’ and ‘Semantically Deviant’ categories. Six sentences being both grammatically and semantically deviant were excluded  from the categories, thus, leaving us with twelve purely grammatically deviant sentences and twenty one purely semantically deviant sentences. In  table 4, the numbers under the  F and % columns represent the number and percentage of teachers who attempted and reconstructed a given sentence correctly. For example,  sentence 1 in the ‘Grammatically Deviant’ category is correctly reconstructed by six teachers out of nine and, thus, its percentage is  66.66.

Table 4 demonstrates that 49.7 % of the teachers succeeded in correctly reconstructing grammatically ill-formed sentences, this is while only 39.14 % of the same teachers correctly reconstructed sematically deviant sentences. It means the obtained results are in line with the teachers’ expectations and contentions who were like-minded in uttering that a semantically deviant sentence is much more difficult to reconstruct than a  grammatically deviant sentencce.

DISCUSSION

The above results embrace general themes that will be expounded in the sections that follow.

Not Confirming the Supportive Role of L1

Corder (1981) hypothesized that acquaintance with the learners and, in particular, their native language plays a crucial part in the process of interpreting and reconstructing a deformed sentence. He holds the view that in case of not having direct access to student writers for the authoritative interpretation and reconstruction of idiosyncratic utterances, we have to recourse to the students’ native language so as to place a plausible interpretation upon it in the conext. Based on the above hypothesis, he argues, if the native language of the writer is not known, the reconstruction or interpretation of the problematic sentence should be suspended temporarily and postponed to later times until we have learned something about the idiosyncatic dialect of the writer. However, if the native language is known, by the process of literal translation we can arrive at the interpreting the sentence plausibly. Then, by translating the native language sentence back into well-formed target language, we can have available the reconstructed sentence.

The researchers, respecting and holding Corder’s hypothesis in high regard, provided the teachers with some general information about the student writers such as their first language (L1), the unversities they attended to, their proficiency level, the courses they had taken in order to find out whether this general information would be of any help for the teachers in unmasking students’ intention from deviant sentences. The results are indicative of the fact that this information proved to be of no or little avail, for the teachers failed to interpret plausibly more than sixty percent of students’ utterances despite the fact that they were familiar with the students’ native language.

 The results of this study confim the results of a study undertaken by Hamid ( 2007) in which he sought to compare the ability of teachers from disparate backgrounds in interpreting students’ deviant utterances. He jumpted to the conclusion that there was no considerable difference in the ability of native and non- native teachers to interpret the students errors. The assumption implicit in the above hypothesis is that all the writer’s idiosyncratic utterances or intended meaning can be plausibly reconstructed or guessed at by having an eye on cross language interference, this is while not all the deformed sentences have their root in the writer’s first language interference but rather, as it is the case with writers in this study, intralingual transfer holds accountable to an overwhelming majority of student writers deviant forms.

It is worth mentioning that Coder’s hypothesis may be of effect or success in environments such as an EFL where students’ native language has dominance over target language and when the writers are not proficient enough in the target language. This hypothesis effect, therefore, wears away in ESL environments where target language and native language stand side by side and when the writers are proficient enough in L2 to overcome the habits of native language.

The Role of Context Downplayed

This study consisted of two phases. In the first phase, the teachers were given context bound deviant and erroneous sentences and in the second phase, the same erroneous sentences, detached from their immediate linguistic context, are presented to the same teachers to unearth their intention via reconstructing them. But the results came as a surprise and were certainly at odds with participating teachers’ expectations who categorically and harmoniously replied positively, while being interviewed, to the role of context in the intelligibility of deviant sentences. This is because there was no significant difference between  the performance of teachers who embarked on reconstructing context-bound or context-free deviant sentences performance of context-bound and context-free teachers.

The results of this study corroborate those of Khalil (1985) who sought to gain an understanding of native speaker teachers’ ability to pass judgment on intelligibility and naturalness of sentences produced by non- native speekers (NNS). In his study a number of grammatically and semantically impaired sentences were presented to the native speaker (NS) teachers both in context and out of context. The obtained results underestimated the role of context. He found that erroneous utterances were rated more comprehensible when they did not accompany the immediate linguistic context. In bald term, the existence of context did not influence native speakers’ ability to interpret the writer’s intent.

Juxtaposing the results of this study with those of Khalil’s study, one may discern a kind of similarity (i.e. context does not have a facilitative role). Great care should be taken not to be deceived by the apparent superficial similarity.   This is because the situation, and, in particular, the nature of activity involved in both studies although their similarities are quite different. In Khalil’s study the respondents were just asked to indicate the degree to which they think  they make sense of the deviant sentences. They were not asked to indicate their understanding by reconstructing the erroneous sentences. Just telling that the sentence is intelligible or not does not guarantee that the he or she has understood the writer’s intended meaning, for he or she might be wrong in his/her hunches.

Upholding this idea, Khalil (1985, p. 345) categorically states that “the intelligibility judgments obtained should not be assumed to reflect native speakers’ actual understanding of the writer’s intent. That is to say, these judgments are not predictors of utterance interpretability”.  By the same token, many a time a teacher thinking that he makes sense of the student’s intention, reconstructs the deviant sentence, the result being that the imposed recontruction disrupts the coherence of the draft. The incoherence brought about as a result of this intrusion functions as a signpost signalling the teacher’s inability in correctly guessing the learner’s intent.

Coming back to the issue of context, we should make it clear that whenever we speak of context, we mean not only the immediate linguistic context but also the situational context. The linguistic context of a sentence entails not only immediate preceeding and following sentences and  paragraphs but rather the whole discourse serves as an enabling factor in rendering help for the reader to put a plausible interpretation on that particular sentence. Khalil (1985, p. 347; see Corder, 1981; Chastain, 1980 ) contends that “research in pragmatics and discourse analysis has shown how the interpretation of an utterance may depend on linguistic clues found in the surrounding utterances -- the universe of discourse for that utterance”. Corder (1981) aptly averred that we should envisage the whole universe of discourse into consideration while attempting to interpret a given sentence. Guided by this knowledge, this prerequisite was materialized when participating teachers were eqipped with students’ write-ups to make use of context in the process of meaning discovery.

In a similar vein, Chastain (1980) states that the more one has a full understanding of the context and the universe of discourse, the more odds are that he/she will comprehend the writer’s intent. The role of the context in the interpretation of utterances has been so unquestionable that few studies attempted to inquire its effect. All linguists and nonlinguists are like-minded that the existence of accompanying immediate linguistic context is incontestable  for the interpretation of utterances. However, this study in which the intersentential context is viewed as an independent variable yields precursory evidence that the accompanying immediate linguistic context does not influence the teachers’ ability to reconstruct or interpret the writers’ intentions contained in the idiosyncratic utterances and yields results that fly in the face of all-agreed upon fact.

This pattern of little or no association between context and intelligibility evokes the query of ‘what was wrong with students’ write-ups in this study that undo the effect of context?’ Or better to say, ‘what features should a piece of discourse possess to qualify it for the context effect?’

A close inspection and scrutiny of the write- ups of the student writers of this study unveils that the discourse produced is not knitted well to qualify it as a coherent discourse. The discourse produced seems more like being composesd of isolated sentences than a unified whole. Even through a cursory inspection, one can find a number of impertinent sentences penetrated into the students’ write-ups. Intrusion of a good number of isolated and irrelevant sentences are enough to offset the influence of context in the interpretation of sentences and render a coherent text as a disintegrated unit in which textual constituents such as sentences or paragraphs are apparetly placed together without being coherently and cohesively tied up together. The point in case is well championed by Khalil (1985, p. 347 ) who averrs that:

coherence of the discourse needs to be considered in analyzing the contribution of context to the interpretation of utterances. The discourse of non- native speakers may be lacking in coherence; such context would therefore  not contribute to successful interpretation of meaning to the same extent as might coherent  discourse produced by native speakers.

One further observation of this study is that we, researchers, and the participating teachers ascribed unconditional role to the context as a facilitating factor but being  ignorant of the fact that context works under conditions that the produced piece of discourse is a unified whole. Khalil (1985) claims that not only the quantity of context but also the quality of context in which the utterances are embeded are of paramount importance. To put it differently,  in order for a context to be of effect  in the interpretation of a given sentence, the surrounding utterances should not only be in abundance but also they should form a well- knitted whole.

Semantically Deviant Sentences Harder to Reconstruct than Grammatically Deviant Ones

The results of this study revealed that teachers found reconstruction of lexically deviant  sentences more difficult compared to their grammatical counterparts. The degree of success for the teachers as regards the reconstruction of  semantically and grammatically deviant sentences was 49.7 % and 39.14 % respectively. This consequence was not unexpected and it is well mirrored in the perspectives and views of teachers being interviewed.

The results of this study also give support to the findings of earlier studies. Chastain (1980), for instance, in his study of native speakers reactions to students errors asserted that intelligiblity is severely blocked by the use of wrong word or the addition and omission of words. The forms of the words, he reasoned, seem to be of a lesser degree of significance in the meaning negotiation process than the proper use of the words. Santos (1988) acceding to the viewpoints of Chastain, on the one hand, and supporting the findings of this study, on the other hand, says that it is precisely this type of error (i.e. lexical error) that “language impinges directly on content. When the wrong word is used, the meaning is very likely to be obscured” (p. 84).

An example sentence from the write-ups of the students of this qualitative study makes the point in case crystal clear. The sentence is: ‘Do you have any idea that access to the information of others without profit how much fun it can be?’

Considering the plausible reconstruction of this sentence, it becomes conspicuous that none of the nine teachers successfully attempted the sentence. The authoritative reconstruction of this sentence (i.e. Do you have any idea what fun it would be having access to the information of which others are denied?) makes it clear that the problem lies in the omission of ‘which’ and improper use of the word ‘profit’, thus, endorsing Chastain’s and Santos’s contention.

The findings of this study also uphold Khalil’s (1985) study of native speakers’ evaluation and interpretation of Arab EFL learners’ written errors and several others. He found that utterances containing grammatical errors were corrected and  interpreted with much accuracy and judged to be more intelligible compared to utterances containing semantic errors that were corrected and interpreted less accurately and judged to be less intelligible. In plain terms, semantic errors, he argued,  were more likely to reduce the intelligibility of utterances than were grammatical errors.

Text Appropriation

Teachers customarily give feedback, or write comments on students’ papers to revise their writing, and, at times, they embark on reconstructing and providing the correct form of the segment of the discourse that they feel needs repairing. They do not seem to have had much success in this undertaking, however. This means that many a time teachers make interpretations and reconstructions of students’ idiosyncratic utterances based on their hunches only. As Hendrickson (1978 cited in Oladejo, 1993) points out, most answers  provided by teachers and linguists to the issues of errors and error correction are conjectural and non-experimental. The results being that teachers write comments or reconstruct the sentences in such a way that prompt the students to relinquish their purpose in order to follow the teacher’s purpose or line of thought. The consequence of teachers’ imposing their own ideas on student writers is that they (students) come to conclude that what their teachers want them to say outweighs what they themselves wish to say. This phenomenon is termed as ‘text appropriation’, ‘text authorship or ownership usurping’ (Brannon & Knoblauch, 1982; Greenhalgh, 1992; Neal et al., 2007; Sommers, 1980, 1981).

As teachers work laboriously at fixing, classifying and providing correct answers for students, or even writing nearly the whole paragraph or the whole piece for students, they are arrogating the students’ right to learn for themselves and denying them of the chance to develop self-editing skills (Lee, 2009). Wonderfully, in classroom writing situations, the reader (i.e. teacher)  uzurps the primary control of writer’s choices, enjoying absolute freedom to amend those choices any time a novice is off the track from the teacher-reader's assumption of what the emerging text ought to look like (Brannon & Knoblauch, 1982). Hence, the teacher, more than the student, determines what the writing will be about, the form it will take, and the criteria that will determine its success.

Radecki and Swales (1988), while interviewing a student on the value of teachers’ comments, discovered an instance of text appropriation by teachers’ comments:

I think what she (teacher) is trying to prove with her comments is that just tells me that I’m off the track… I’m not explaining my subject, I’m not explaining what I want to say and she’s really stressing on that… She doesn’t care what I’m writing about or what I’m doing unless she sees her own topics and format in my paper. (p. 362)

CONCLUSION AND IMPLICATIONS

Teachers, assuming themselves as authorities in the classroom setting, grant themselves unlimited right to intervene at any time they feel students go awry from teacher’s ideal text. Teachers, more often than not, mistakenly presuming or taking for granted that they can understand students’ intentions embedded in the erroneous and idiosyncratic utterances, embark on reconstructing students’ deviant sentences just based on their intuition or sheer hunches. So doing, they not only disrupt the unity of the students’ produced discourse but also usurp the student-writer’s voice and train their effort and thought towards meeting their own purpose in lieu of students purpose in writing. The research results are indicative of the fact that teachers’ reconstructions or interpretations of learners’ intent enclosed in the erroneous utterances are not always free from errors and hence unreliable. It is not unlikely that teachers may impose their self-perceived subjective interpretations in lieu of actually discovering learners’ intentions in idiosyncratic utterances.

Spandel (2005) utters a word of caution against usurping the student-writers’ voice and informs us of the significance of honoring the rights of student writers, such as the right to go off the topic and to write badly. To free themselves from the arduous and drudgery task of marking student writing, teachers can respect student writing wearing new glasses, treating it as an artifact produced and possessed by the student writer (rather than the teacher) and showing it greater respect. Conceding to Spandel’s viewpoints, Hamid (2007) categorically contends that teachers who provide written feedback to students and often reconstruct their idiosyncratic constructions in their absence need to practice caution and sensitivity to make sure that they do not impose their own meanings on students’ writing and thus colonize their thoughts. Rather than labeling student writers’ drafts as poor writing or replenishing it with red ink, teachers should figure out what the writing tells them about students’ personal perspectives, what their abilities or debilities in writing are, their worldviews, their idiosyncrasies and, above all, who they are as people and as writers (Murray, 1985, cited in Spandel, 2005).

Greenhalgh (1992), similarly, argues that when teachers use their responses to guide or superintend writing, they are, in fact, hold captive by the voice of the student writer. Instead, teachers should encourage students to take control of their own writing. “To do this, teachers must take note of shifts in their own voice in their responses, talk to students about the power dynamic in response, and help students hear their own voices as they take or relinquish control of their writing” (Greenhalgh, 1992, cited in Neal et al., 2007, p. 64). 

The findings of this study did not uphold Corder’s hypothesis maintaining that teachers being acquainted with students’ native language are better off or in a vantage point of conjecturing exact or near exact intent of students enclosed in idiosyncratic utterances. Yet, another conclusion is that not any context of a deviant and erroneous utterance conduces to the reconstruction or interpretation of that utterance. Not only should the context of a deviant utterance be rich enough in terms of quantity of utterances preceding or following that utterance, but it should also be coherent in terms of its quality. Khalil (1985) contends that the discourse of non-native speakers may be devoid of coherence; as such it would not contribute to successful interpretation of meaning as much as the coherent discourse produced by native speakers.

Given the small corpus of this study, great care should be taken in generalizing its findings. Additional research is required to validate and verify its findings. This deficiency notwithstanding, the study has the potential of providing teachers and researchers with insightful look into the nature of error correction and feedback.

The implications of these findings for the teaching of writing, especially, in the EFL settings are quite clear. When meaning negotiation is the purpose of writing, student writers should be made wary of the role of linguistic context as well as the significance of semantic and  formal features of language. It is, therefore, incmubent on writing  instructors to lay due stress on the development of these skills  in the writing process and in the sequencing of materials so that student writers could effectively communicate their meaning. Rather than acting on behalf of students by re-writing student texts (and altering their meanings) mistakenly (pretending to be extremely devoted and committed), writing teachers could talk to students to find out what exactly they want to say.