Institutional Integrated Teletandem : What have we been learning about writing and peer feedback ?

We defi ne institutionally integrated teletandem (iiTTD) as a series of teletandem sessions that are embedded in regular foreign language lessons, so that such lessons both feed and are fed by teletandem practice. In this paper, we analyze prospective Brazilian teachers correcting American students’ writing productions in Portuguese during their participation in an iiTTD. We address issues about how pre-service FL teachers (participants in iiTTD) offer writing feedback to their peers. Data included all the texts written and revised during the length of their partnership. Results show that most participants tend to correct their iiTTD partners’ written productions in a direct way, focusing mainly on language form and according to their beliefs and experience as learners. Key-words: teletandem; writing process; corrective feedback; foreign language learning. Solange Aranha, Suzi Marques Spatti Cavalari

Within this context, the participants are responsible for the activities carried out in order to improve their performance in the foreign language (FL) they are learning and are responsible for fi nding ways of teaching their own language.Besides, they are responsible for managing their schedule, the time and the activities in each language.The process and the product of this interaction are supposed to be neither graded nor part of their language syllabus.Aranha and Cavalari (2014) describe a modality of teletandem developed at UNESP -São José do Rio Preto in which the teletandem practice is incorporated into the language syllabus and a series of mandatory activities are proposed.This modality is named institutional-integrated teletandem (iiTTD) (as detailed described in Aranha;Cavalari, 2014 and Cavalari;Aranha, ongoing research).In general terms, the differences can be pinpointed as follows: Table 1 -Teletandem and Institutional integrated Teletandem: contextual differences (based on Luvizari-Murad, 2014:8) TTD iiTTD

Time and place of interactions (sessions)
Partners choose the most convenient time and place.
Sessions always happen at the university Teletandem lab within class schedule both in Brazil and the United States.

hemes used during participants´ conversations
Participants are responsible for deciding whether they want to establish a specifi c theme beforehand.
Sessions are supposed to begin with the feedback of revised texts.After doing so, participants may choose whether to continue on the text topic or change topics.

Tasks to be developed prior to sessions
None, unless participants decide to do so.
Written production about themes related to the FL class discussion.Participant A sends a text to his/her partner prior to the session.The partner should correct and comment the text in various levels.The beginning of the next session is supposed to be devoted for discussing the corrected version in terms of form and content.

Process assessment and feedback
Assessment and feedback are only provided if participants agree to do so.
Participants are graded on iiTTD activities.

Particpants´ support
The participants may, if they wish, have help from a teacher or graduate student.
Participants should report to their language professor (and grad students who help develop the activity).

Data collection
Individually carried out if the partner is involved in research.
Collectively carried out.All participants are involved in collecting data after each session in the iiTTD lab.Computers are specially fed with recording programs.
Within this modality, one of the mandatory activities, the writing exchange, which serves as a trigger for the following session (Aranha;Cavalari, 2014), is far from having shared instructions among the groups enrolled in the project.Each teacher, responsible for a specifi c group, has his/her own way of approaching the instructions students are given, so that the draft product -the text to be used for correction -is created.By the same token, the feedback students are supposed to provide is also dependable on the teacher's approach or, ultimately, on the students´ themselves.
In this modality, writing feedback is made solely in pairs, i.e., the teacher does not interfere or participate directly in the process.He/she is responsible for grading the fi nal product, but not for intervening in the correction.This paper addresses issues about how Brazilian FL teachers-to-be participants in iiTTD offer writing feedback to their peers 1 , specifi cally focusing on the types of corrective feedback given to texts written in Portuguese by American students.

Theoretical framework: Feedback to writing
Much has been discussed about response to students' writing (Zamel, 1985;Freedman et al., 1987;Saito, 1994;Truscott, 1996;Lee, 1997;Storch, 2005, Figueiredo, 2005a;2005b;Bitchener et al., 2005;Hyland;Hyland, 2006;Wigglesworth;Storch, 2012;among others) and some of these studies propose that it can be classifi ed according to the source of the response: • teacher feedback: (i) teacher's correction of students errors; (ii) teacher-student conferencing (sessions in which a teacher and a student talk and work together on a piece of writing); • peer feedback: (i) a student provides feedback to another's preliminary draft; (ii) peer conferencing (peers discuss teacher's feedback in pairs); • self-correction (or self-assessment): a student reads and examines his/her own writing.
Different approaches have been used to investigate each of these aspects of feedback provision to FL writing.Peer feedback, which is the focus of this paper, is usually investigated in terms of peer conferencing, but there seems to be a gap in the study of the types of 1. See Consolo; Furtoso (in this issue) for a discussion on other elements of assessment in the teletandem context.corrective feedback on a student-student basis 2 .For this reason, the categories presented in this section are based on studies conducted on a teacher-student basis.
As to teacher feedback, most research (Zamel, 1985;Freedman, 1987;Saito, 1994;Bitchener et al., 2005;Hyland;Hyland, 2006) focuses on (i) the method or the strategies used by the teacher (direct/ explicit vs. indirect/implicit strategies; with or without codes); (ii) the focus of feedback (focus on different language aspects vs. on content); or (iii) students' perceptions of teacher feedback.According to Bitchener et al. (2005), direct or explicit feedback occurs when the teacher identifi es an error and provides the correct form, while indirect strategies refer to situations in which the teacher indicates that there is an error but does not provide a correction, leaving the work to be done by the learner.In the case of indirect feedback, the teacher may indicate the type of error with a code or by means of comments.These authors, nevertheless, do not address the use of technological tools that can be used for feedback correction (such as the "track changes tool" by Micrososft).
It is important to note that there has been a great debate in the SLA research area (Truscott, 1996;Lee, 1996;Ferris, 1999;Bitchener et al., 2005; among others) whether providing corrective feedback on second language writing could lead to either short or long-term improvement.Bitchener et al. (2005) offer a review on the most relevant studies involved in such debate and the authors report that (i) providing feedback (either direct or indirect) seems to be more effective than providing no feedback at all; (ii) both direct and indirect feedback strategies lead to similar levels of accuracy over time.However, the authors point to the fact that there is not much research on the effect of other feedback strategies, such as teacher-student conferencing, peer-editing sessions or students' error logs (cf.: Bitchener et al. 2005: 194).Some of the studies conducted in this gapped area are those of Figueiredo (2005a;2005b) and Wigglesworth and Storch (2012), who investigate feedback to writing within a sociocultural framework.
2. In the scope of iiTTD studies, Brocco's thesis (2014) also focused on the types of corrective feedback provided.Figueiredo (2005b) analyses self-correction in relation to collaborative corrective strategies, i.e., teacher-student conferencing and peerfeedback.This study considers peer-feedback in relation to the kinds of errors peers correct and learner´s perceptions on this type of feedback.The author concludes that the collaborative strategies seem to be not only more effective, but also more valued by learners.Wigglesworth and Storch (2012) investigate students collaborating both in writing and in processing writing feedback (peer conferencing).In the study, the actual corrective feedback is given by the teacher on the assumption that it will lead not only to improved accuracy in the short term (on immediate redrafts) but to language learning in the long term.However, the authors remark that for feedback to be effective in terms of language learning "it needs to be processed and acted upon" (Wigglesworth and Storch, 2012 : 368).Feedback processing, according to the authors, involves learners noticing and working on the feedback, i.e., acting in the process to implement the product.Their study focuses on how (or if) working predominantly in pairs, and receiving feedback on their writing in pairs, can enhance the language learning opportunities for learners through their ability to scaffold each other's contributions and knowledge.Basically, they describe a student-student conferencing modality as a collaborative way to process feedback and conclude that this kind of collaboration can enhance learning by providing opportunities for the discussion of language.
According to Lee's definition (1995), peer feedback can be characterized as "a process where students read drafts written by their fellow students and give each other suggestions to improve the writing" (Lee 1995:58).Since students write their texts to a real audience and not to a pseudo-reader (the teacher), the author argues that "peer reviews refl ect writing as a truly communicative process rather than an artifi cial, lonely exercise" (Lee 1995:59).
Institutional integrated teletandem modality (Aranha;Cavalari, 2014) offers opportunities both for writing practice in the participants' target language and for text reviewing in their native language (or the language they feel competent in) 3 .From a sociocultural theoretical 3. Telles (2006) considers that teletandem participants do not have to be natives in the language they help their partner learn.
perspective, the writing process in this context can be considered collaborative, if we take into account that students (i) receive some input in their regular FL lesson before they start writing; (ii) have their written production corrected by their iiTTD partner; (iii) can discuss such correction during iiTTD session and (iv) may rewrite the text according to the received feedback (Cavalari; Aranha, ongoing research).This modality, therefore, encourages a collaborative writing process once the tasks are conceived as integrated into the foreign language pedagogy.It should be noted, however, that the writing feedback (concerning focus and strategy) within this process is performed by the peers according to their experiences and beliefs.
Considering these issues on collaborative feedback strategies, in this paper, we focus on the feedback offered by peers in iiTTD writing task.It is necessary to remark that, differently from the contexts investigated by the studies on peer feedback mentioned in this paper (which focus on students who speak the same mother tongue and give feedback on texts in their target language), in iiTTD, the peer who provides feedback is a native (or a competent speaker) of the language in which the text is written.

Methodology
The process of data collection and fi ling is thoroughly described in Aranha, Luvizari-Murad and Moreno (forthcoming), and, for the purposes of this article, i. e., analyze the written production and feedback, only this aspect of data collection may be summed up as follows: 1. Students are required to upload on the online learning platform -TelEduc (http://prograd.ead.unesp.br) the fi rst text that was sent to their partner, the corrected draft, and the rewritten version 2. The project monitors are responsible for transferring all the stored data to an external hard disk (HD).Each fi le is named by means of a shared code that indicates (i) the number of the written text to be stored; (ii) the version of data in question: fi rst draft, revised or rewritten; (iii) the number of the participant (that corresponds to the number of the computer he used); (iv) and the group specifi cations.For instance, the code (3REDOp1QUI) corresponds to the third original text in Portuguese saved by the user of computer 1, who belongs to the Thursday 4 iiTTD group 5 .
Data used in this study were collected in the fi rst semester of 2012 within a group of 17 students who were attending English as foreign language classes in the second year of a Language and Literature teaching major at UNESP -São José do Rio Preto.These students participated in iiTTD from March to May as part of their course syllabus.Eleven out of these 17 students signed the Letter of Consent and uploaded the revised versions of the texts in Portuguese in TEleduc.Two out of these eleven students did not write a text in accordance with the American teacher's instructions 6 .Therefore, the nine participants considered in the analysis are ( 01), ( 03), ( 04), ( 06), ( 07), ( 09), ( 11), ( 14), ( 19).
Different studies on teacher corrective feedback (Zamel, 1985;Freedman, 1987;Saito, 1994;Bitchener et al. 2005;Hyland, 2006) acknowledge two broad categories of analysis: one concerning the focus and one concerning the strategy or approach adopted by the teacher.Although these categories refer to the teacher pedagogical practice, it may argued that they are adequate for this analysis once the Brazilian participants are teachers-to-be.Hence, the results presented in section four were analyzed according to the following categories: A. Feedback strategy: direct vs. indirect; with or without comments and/or codes.In this category, the technological tool used to offer feedback is also considered because corrections are made by means of a word processing software (Microsoft Word) which offers collaborative tools (such as "track changes" and "comments").These tools might have an impact on the strategy adopted, once the changes may be seen by the writer.
B. Feedback focus: different language aspects vs. content.
4. QUI corresponds to Quinta-feira, which means Thursday.REDOp means "fi rst draft" in Portuguese. 5.This data bank provides a letter of consent that is signed by the participants who agree with providing data for research.6. Teacher's instruction: Watch this TV commercial (http://www.youtube.com/watch?v=X_RAF130AGA) and write a letter to "your future self".Imagine you will read this letter in 20 years from now.

Results and discussion
This section presents the results of investigating the types of corrective feedback provided by nine Brazilian participants in iiTTD to texts written in PFL (Portuguese as a Foreign Language) by Americans.The results are presented according to the feedback strategy (including the technological tool used) and to the feedback focus (including the nature of comments when there are any).
Out of nine participants who wrote the text according to their teacher´s instructions, one participant (06) used a direct strategy without comments, i.e., he revised his/her partner´s text by substituting the error with the form he considered correct and by using the "track changes" tool.Example 01 illustrates the use of track changes 7 Example 01 -participant 06 It should be noted that using a word processor software (Microsoft Word, in this case) facilitates the revision since the elements the Brazilian participant deletes are stricken, but remain in the text (semester).Nevertheless, this corrective procedure can be characterized as prescriptive and focused on the product, since it does not provide options for the learner to make decisions on his/her own writing process.The Brazilian's conception of writing and corrective feedback seems to be in opposition to Lee's defi nition of peer feedback who suggests that peer feedback is a "truly communicative process" (Lee 7. I hope you can still read Portuguese because now I love this language!This semester will be my last semester of Portuguese because I will not have space for this lesson.I am sad about that because I have learned so much about Portuguese and I am not prepared to fi nish.1995:59).One might argue that participants are not given explicit instructions about writing feedback procedures in iiTTD.However, this participant's prescriptive correction cannot be ignored as one of the possible strategies students might use.
Five other participants (01, 04, 07, 08, 14) used this same direct strategy, i.e., the use of track changes, but they added notes to some of the corrections by using the "comments" tool.Four participants made their comments in Portuguese (04, 07, 08, 14) and one of them, in English ( 01).Let us turn now to participant 07 remarks, which illustrates this category with the comments in Portuguese: Example 02 -participant 07 As it can be seen, participant 07 (in the example) does not comment all the corrections made and neither do the other four participants who use the comments tool.It may be inferred that leaving some corrections without any comments might be caused by the fact that participants know they will have the chance to talk about the revision during the iiTTD session.However, there is no evidence to whether this is really the reason why they do not add notes to some of the corrections or to which criteria they adopt to decide which corrections deserve commenting or not.It can also be noticed that the nature of the comment in (07) 8 tries to acknowledge the writer's background knowledge in Portuguese (Remember, it is feminine).The comment here sounds like a reminder of what the writer is allegedly supposed to know and, thus, reveals an attempt of a collaborative approach to correction.
8. Remember, it is the feminine because it is about "things", so use the feminine.And if the pronoun is the object, then you cannot use "they", but him/her.Correct form: "let them".
Similar to this strategy, but using different tools, is the one adopted by participant 19, who uses the "highlight" tool to draw attention to both error (highlighted in red) and correct form (highlighted in yellow) 9 .Example 03 illustrates this participant's strategy: Besides, this participant adds comments by using "endnotes" as the following example (related to example 03) demonstrates: As it can be observed, this comment focuses on form (the use of subjunctive) and aims at explaining the error by using metalanguage in a prescriptive way.It might be said that participant 19, who is a language teacher-to-be, believes this is an appropriate way of correcting and may lead to effective learning.Nevertheless, he does not take into account that his/her partner might not be in the language fi eld, which may not foster collaborative communication in the iiTTD session.On the contrary, depending on the partner´s personality or expectations, it may actually impair further communication.
Although participant (11) also uses a direct strategy, he uses other tools and his comments are of a different nature.He/She changes 9.I hope that in 20 years, I will have a career that I like as well as fi nancial stability.I hope I can give my son as much attention as he deserves… the color of the font to indicate the error and the correction and adds the correct form between parentheses.Example 04 10 illustrates this strategy: Example 05 -participant 11 In cases he/she wants to insert a comment, it is "embedded" in the text, between parentheses, in a different color, as it can be seen in example 05: Example 06 -participant 11 The corrections focus on form, yet the nature of the comment 11 is collaborative and, in this case, related to sociocultural issues, i.e., how a more appropriate word to the context (namorando/dating vs. saindo/going out) would fi t the writer´s needs or intents according to the participant who is correcting.The rise of linguistic and cultural issues is coherent with the development of the tasks in iiTTD, presented in table 01 in relation to "tasks to be developed prior to sessions".10.You were a horrible boss, but you wanted to improve.You liked fashion, dogs, music, volunteer work and going out with your friends.You also liked painting, but you were so busy that you did not paint very often.11.You were going out (the term "was going out" is grammatically correct but it refers to relationships which ar not very serious.If you want to refer to a serious relationship, use "dating") Participant (01), in example 06 12 , also makes comments of a collaborative nature and makes direct feedback corrections.The comment (I will explain that to you after (sic)) reveals that this very participant, similarly to participant (11), is trying to follow the procedures of iiTTD when he indicates that the beginning of the following session will be "devoted for discussing the corrected version" (as also shown in table 01 in this paper).
Example 07 -participant 01 This corpus presents just one sample with the use of direct and indirect strategies.Like all the others, participant (09) uses direct strategies, but he also corrects in an indirect way by highlighting some words (which are neither corrected nor commented), as example 07 shows: Example 08 -participant 09 12.I hope you fi nd a career you love.Perhaps, you will be in a University in Brazil or the United States teaching History.
In this case, it may be assumed that the highlighted word (mundano) 13 may be left to be discussed during the iiTTD session.Although these data offer no evidence to confi rm this assumption, the highlighted word seems to refer to the contents of the text, i.e., a piece of information that does not make sense.The nature of the comments 14 , though, are not collaborative and do not acknowledge the writer´s background knowledge.They just provide the right form, sometimes with the use of metalanguage.It is interesting to observe that the use of the collaborative tools (29 comments are added to this text) do not make this participant's correction collaborative, but the fact that he might have left some words (possibly related to content) to be discussed later could refl ect his/her perception of which aspects are more relevantly treated during the interaction.
According to the data analysis presented, most corrections made to the nine texts were direct and focused on the form, especially concerning: • spelling (semester/semestre) -in example 01; • preposition (the regency would be espaço para and not pela) -in example 01; • vocabulary adequacy (aula to mean class, instead of classe which would constitute of a literal translation) -example 01; • verb form (tense and mood and use of refl exive verbs) -examples 03 and 06; • gender (pronouns that refer to feminine nouns) -examples 02 and 07; • stress (ultimo/último) -in example 01.
• number concordance (pronouns that refer to plural nouns)example 02.
13. Anyway, the closure of each situation is almost totally dependent on the way you react to it.If you fi nd a positive side in everything, you can resist any storm-it doesn't matter how hard or mundane can think it is.14.The pronoun must be feminine, because....
Eight out of nine participants make comments (either using the comments tool or not) and four (01, 07, 09, 11) of these eight evidence some attempts of a collaborative approach to correction, which, according to Figueiredo (2005b), seem to be more effective and valued by learners.The other participants' comments are all prescriptive in nature and tend to view the text as a fi nal product.Based on the categories proposed in section 3.3, the analysis shows that the corrective feedback provided by the Brazilian participants can be summarized by the following diagrams: As the diagrams show, most corrections are direct and focused on form: seven participants (01, 03, 04, 06, 07, 14, 19) use a direct strategy with variations in the approach (comments an codes) and the tool; one participant (11) uses a direct strategy and focuses both on form and on content; and one (09) uses both direct and indirect strategies.It can be observed that, even though corrections are mediated by collaboration tools (track changes and comments), only four out of nine Brazilians try to use collaborative feedback strategies (in a few instances).In most corrections, participants do not seem to acknowledge (i) the need for the partner to work on the feedback given and (ii) the opportunity to discuss the feedback on iiTTD sessions.This situation is not coherent with iiTTD proposal (table 01 in this paper) and contrary to Wigglesworth and Storch's (2012) fi ndings, i.e., for feedback to be effective in terms of language learning "it needs to be processed or acted upon".

Final remarks
By focusing on one of the aspects of the writing process (the corrective feedback given by Brazilian participants teachers-to-be), this paper is an attempt to point out the potential of this context to develop writing skills, to assist FL teacher education courses and to implement research in Applied Linguistics.
The analysis shows that most participants apparently tend to correct their iiTTD partners' written productions according to their beliefs and experience as learners, coherently with the feedback usually provided by teachers, i.e., by eliminating the error, providing prescriptive correct forms and preventing students from exercising their own learning strategies.It may be said that this approach to correction is not in accordance with a pedagogical context that aims at promoting autonomous and collaborative foreign language learning, as the one described in this paper.This situation suggests (i) a gap on these participants' (teachers-to-be) foreign language education concerning written feedback provision and (ii) the need of including instructions about how to provide feedback in iiTTD in order to be in alignment with the purposes of this context.Nevertheless, it should be acknowledged that these participants certainly did the best they could in this new and challenging paradigm of autonomous and collaborative language learning.This study also shows that further investigation on (i) the types of errors present in the texts and corrected by the iiTTD participant and (ii) the accuracy of corrections may contribute for course designers in the development of Portuguese as a foreign language courses.Finally, it must be remarked that this new teletandem modality has generated an enormous quantity of data that can be studied and analyzed from diverse methodologies and theoretical approaches.As far as writing is concerned, the research group at UNESP -São José do Rio Preto is focusing on: types of errors, genre instructions for writing, transcultural issues, peer conferencing (discussing the corrective feedback in iiTTD sessions), among other issues that can assist FL teaching-learning not only within iiTTD context.

Received in March, 2015
Accepted in August, 2015 E-mails: solangea@ibilce.unesp.brsmscavalari@gmail.com (A) Direct feedback using different strategies and with different foci (8 participants) Diagram 01 -Direct feedback using different strategies and with different foci (B) Direct and indirect feedback using comments and with focus on form (1 participant) Diagram 02 -Direct and indirect feedback using comments and with focus on form