Second language learners ’ reflections on the effectiveness of dictogloss : A multi-sectional , multi-level analysis

Despite the extensive research conducted regarding Focus on Form instruction, no conclusive results have been provided concerning (a) the issue of which techniques contribute most effectively to L2 acquisition, and b) at which level of proficiency those techniques should be implemented for best results. Dictogloss, one of these techniques, has been proven to be effective (Fortune, 2005; Kowal & Swain, 1994; Malmqvist, 2005; Nabei, 1996; Swain, 1998). While previous studies evaluating dictogloss explored feedback opportunities and the amount and type of language related episodes produced, fewer studies have reported on the effectiveness and its applicability according to proficiency level (Fortune, 2005; García Mayo, 2002) and none have explored learners’ conceptions about the task. Therefore, this study seeks to determine which proficiency level might be most appropriate for the implementation of dictogloss and to gather learners’ opinions regarding its usefulness and effectiveness. A total of 497 participants enrolled in novice-mid (N = 275) and advanced-low (N = 222) levels took part in the study. All participants engaged in two dictogloss tasks and completed a survey afterwards. Overall, results indicate that dictogloss was better received by advanced-low level students and that most students found it both useful and effective for learning.


Introduction
Research on SLA has witnessed a growing body of studies that address how focus on form (FonF) instruction could enhance L2 acquisition (Doughty, 1991;Doughty & Varela, 1998;Doughty & Williams, 1998;Ellis, 1998;Leeman, Arteagoitia, Fridman, & Doughty, 1995;Long, 1991;Long & Robinson, 1998;Norris & Ortega, 2000;Spada, 1997;etc.).Another widely investigated aspect has been the evaluation of different techniques that can be incorporated into FonF instruction.However, despite the great number of studies conducted, there is no concrete empirical support in favor of one FonF technique.Nevertheless, there is a consensus among researchers that (a) noticing is a prerequisite for acquisition, (b) comprehensive input alone is not sufficient for acquisition, and (c) producing output is necessary for L2 attainment.
FonF instruction can be implemented through a variety of techniques (input enhancement, input flood, output enhancement, etc.).One such technique, known as dictogloss, leads learners to pay attention to form while engaging in text reconstruction.Previous research has yielded differing results concerning the effectiveness of this technique (Fortune, 2005;Kowal & Swain, 1994;Malmqvist, 2005;Nabei, 1996;Swain, 1998).Most studies have explored the opportunities for feedback obtained during the task and the type and number of language related episodes (LREs).LREs are defined as parts of students' dialogs in which they "talk about the language they are producing, question their language use, or correct themselves or others (Swain & Lapkin, 1998, p. 326).However, such studies have paid less attention to the viability of the implementation of dictogloss with learners displaying different degrees of proficiency and, to my knowledge, no studies have focused on exploring learners' opinions and attitudes towards this task and its components.With this in mind, the current study seeks to expand on previous research by exploring students' conceptions regarding the effectiveness of dictogloss as well as the effectiveness of dictogloss with learners of different proficiency levels.

Dictogloss as a focus on form technique
The idea of dictogloss as a collaborative task was first introduced by Wajnryb (1990) and consisted of an adapted type of dictation that involved students working collaboratively to reconstruct a passage that was read to them.The task preparation requires that the instructor select a text, either authentic or created specifically for the purpose of the task.The text generally contains several instances of a targeted form.Shak (2006) suggests that task conductibility and success are determined by (a) the level of language used in the text, which should be compatible with the learners' linguistic ability, and (b) the target language features made apparent to the learners, which could be achieved through recurrences of the features in the text.The texts used in the present study were adjusted to each of the proficiency levels and included target features that were being covered in class at the moment of data collection.
According to Wajnryb, a dictogloss task should have four stages: 1. Preparation: The topic is introduced and the instructor conducts a warm-up activity that contributes to the students' familiarization with the main ideas included in the text.At this time, the instructor should introduce new vocabulary and explain how the activity works in addition to organizing students in small groups or pairs.2. Dictation: The instructor should read the passage twice.The first time students are instructed to listen and the second time they are to take notes.The text is supposed to be read at normal speed, identically both times.3. Reconstruction: Learners share their notes and discuss how to reconstruct the text.The instructor monitors the activity but does not provide any type of input during this stage.4. Analysis and correction: This stage could be adapted to the needs of each classroom.The main goal is to share the reconstructed texts and have students engage in error correction and analysis.One option is for the instructor to write students' texts on the board and provide feedback.This part should be carried out in a sentence-by-sentence manner.The instructor can later provide the original version for students to consult.
Many studies do not follow this procedure exactly.For instance, both Fortune and Thorp (2001) and Fortune (2005) did not require students to attempt to reproduce the dictoglos text, but to produce a text of their own, maintaining the overall original meaning.In those studies, the reconstructed texts may contain alternative lexical and grammatical forms.Shak (2006) proposes an overall different approach.In her version of the task, participants are asked to reconstruct the text individually and then in groups.In Wajnryb's version, participants start off in groups sharing notes and then conduct the reconstruction process with a partner.Allowing learners to start the reconstruction process individually could limit the opportunities for noticing, since it is completed initially without engaging in metatalk, one of the main goals of the dictogloss task.
Another common variation regarding task implementation has been the use of L1.Scott and De la Fuente (2008) maintain that L1 use always occurs to some extent in the foreign language classroom and therefore suggest that L1 prohibition could be futile.They also affirm that exclusive use of the L2 might impose a cogni-tive challenge and might inhibit collaborative interaction, impeding the use of metatalk (p.109).Malmqvist (2005) addressed this issue stating that L1 use was allowed in her study because the participants were beginners or false beginners and they lacked the ability to conduct metatalk in the target language.During the dictogloss sessions carried out in the present study, participants were reminded and encouraged to use the L2.Lastly, the dictogloss was implemented following Wajnryb's (1990) original format, and included all four stages proposed by her.

The effectiveness of dictogloss and metatalk
Previous studies (Fortune, 2005;García Mayo, 2002;Kowal & Swain, 1994;Malmqvist, 2005;Nabei, 1996;Qin, 2008;Salazar Campillo, 2006;Swain, 1998) have shown a variety of results concerning the effectiveness of dictogloss with regard to focusing on form, noticing interlanguage gaps and acquiring the given forms.This section includes an overview of research in which the dictogloss task was evaluated in light of the metatalk produced and measuring its effectiveness.
Within the context of this task, collaboration is thought to generate metatalk, which directs learners' attention towards certain linguistic features through reflection and discussion (Kowal & Swain, 1994;Swain & Lapkin, 1998).Exploring the intricacies of metatalk, Swain (1998) carried out a study with 48 students in an 8th grade French immersion class and sought to determine whether students could engage in metatalk following the modeled example they were previously provided and whether there was a relationship between metatalk and second language learning.There were two groups in her study, the metatalk group (N = 26; it was exposed to modeled metatalk and explicit rule t eac h in g ) an d th e c o n tr o l g r o u p ( N = 2 2 ; it w a s no t e xpo se d t o m o d el ed metatalk and explicit rule teaching).Results indicated that the metatalk group produced 2.5 times more LREs than the control group (metatalk group: 14.8; control group: 5.8), showing that modeling the metatalk increased the production of LREs.Findings also suggested that students' conscious reflection about language might be a source of language learning.Kowal and Swain (1994) conducted their study with 19 students enrolled in an 8th grade class.The students were required to take part in four dictogloss tasks that were designed to provide practice in the use of the French present tense.While carrying out the activity, all of the stages were completed: They discussed vocabulary and topic, they heard and reconstructed the text and some texts were chosen by the teacher for the final discussion.Results showed that form was the focus of the students' discussion and that peer feedback was highly important in moving from semantic processing, required for understanding, to grammatical processing, needed for production.Fortune (2005), whose advanced level participants indicated a higher level of readiness to engage in metatalk in comparison to intermediate level participants, carried out a study in which students at two different levels of proficiency took part in a dictogloss task.The main goal was to investigate the metalinguistic terms used by the learners during interaction.It also sought to compare how frequently metalanguage was used by intermediate and advanced learners to establish whether metalanguage use enables more sustained engagement with the targeted form and whether it helps learners to attend to those forms more readily.Results indicated that advanced level students used metatalk 46.4% of the time and that intermediate level students used metatalk only 29.4% of the time.The advanced level students concentrated more on form, and they employed more metatalk in doing so, showing more readiness than the intermediate level students.However, the researcher indicates that LREs without metalinguistic content do not necessarily hold less value in the production of output.In this case, the task was not evaluated as an instructional technique capable of enhancing L2 acquisition but as a context that provides students with the opportunity to produce metatalk.
With an operationalization similar that of Fortune (2005), Leeser (2004) conducted a study using dictogloss in a Spanish content class.His 42 participants were required to engage in two dictogloss tasks.The first dictogloss maintained a structure fairly similar to the original one proposed by Wajnryb (1990); during the reconstruction, however, the participants were required to say aloud everything that they were writing in order to reflect why they chose certain forms.No details are mentioned about Stage 1 or 4.More significant modifications were introduced during the second dictogloss.First, a review of the topic was presented, and students were also given a handout including Spanish aspectual differences (preterit vs. imperfect) and were allowed to ask questions after going over the handout.They also watched a video showing two participants reconstructing a text and discussing linguistic difficulties that they encountered during this stage.The analysis and correction stage was not completed since the students were dismissed after they finished the reconstruction.Results show that students produced a total of 138 LREs.Of these LREs, 39.86% were lexical and 60.14% were grammatical (and more than a half of them related to subject-verb agreement and tense/aspect choice).Of the linguistic questions that emerged during the reconstruction, 76.81% were solved correctly.As in Fortune (2005), only an analysis of LREs was conducted but no pre-and post-tests were administered.
Finally, with a fairly different interpretation of dictogloss, Qin (2008) attempted to account for possible distractions or off-tracking that students could experience during the completion of the activity.Previous research findings (García Mayo, 2002;Nabei, 1996;Swain, 1998;Williams, 1999) showed that students might not attend to the intended forms; thus, it was both important and fruitful to employ mini-techniques in order to raise awareness and to model the reconstruction (Swain, 1998).In Qin's (2008) study, the stages were organized in a different manner: During stage 1, the instructor first introduced the topic and then handed out copies of the text.Students were instructed to read individually and then discuss the meaning with partners.In addition, the instructor emphasized the targeted form (passive voice) and reminded the participants to pay special attention to it.This modification was introduced in order to prevent students from focusing their attention on a nonintended form.The overall completion of the task, and the opportunities for metatalk, is not circumscribed to facilitate the targeted forms.It provides an arena for ample production of LREs (lexical, discourse, grammatical, etc.) as well as fostering the acquisition of both the targeted form and additional forms and concepts.Qin (2008) conducted the task without including a listening component as the text was not read to the students.Therefore, it should not be considered dictogloss.When engaging in a dictogloss task, the overall completion and the opportunities for metatalk, is not circumscribed to facilitate the targeted forms.It provides an arena for ample production of LREs (lexical, discourse, grammatical, etc.) as well as fostering the acquisition of both the targeted form and additional forms and concepts.Qin (2008) study is reported here, nonetheless, since it suggests the inclusion of modifications to foster awareness-raising and noticing during metatalk.
In sum, previous studies investigating dictogloss have mainly focused on the number of LREs that learners produce during metatalk, the terminology used during interaction, the quality and effects of metatalk, and the opportunities for feedback that dictogloss provided.However, to my knowledge there are no studies that have taken into account students' impressions of the dictogloss activity or researched students' opinions regarding the effectiveness of metatalk.Therefore, the present study intends to incorporate a unique element into the discussion of the effectiveness of this task.Additionally, it has been suggested that metalinguistic activity can be performed both explicitly and implicitly (García Mayo, 2002;Gutiérrez, 2008); therefore, the present study seeks to identify whether learners can recognize their reflections about the language even when not discussing rules explicitly.Lastly, since learners in novice level classes might not be cognitively ready to fully engage in metatalk and comprehend how all the stages of the task serve the purpose of focusing on form (Fortune, 2005;García Mayo, 2002;Leeser, 2004), the present study also explores the feasibility of its application as perceived by learners at different levels.To address the aforementioned issues, the following research questions guided the investigation: 1. Do advanced-low and novice-mid level students of L2 Spanish find the dictogloss task conducive to learning?2. Do advanced-low and novice-mid level students of L2 Spanish identify the acts of talking and thinking about the language as part of the dictogloss task? 3. Do advanced-low and novice-mid level students of L2 Spanish recognize metatalk as an effective practice that enhances acquisition?

Method
This study sought to gather attitudes and impressions concerning the effectiveness of dictogloss from students in L2 Spanish classes at two different levels at two different American universities.Participants in both institutions engaged in two dictogloss sessions and completed a survey afterwards.The data obtained from both institutions were compared and contrasted in order to respond to the research questions.

Participants
A total of 497 learners of Spanish participated in the study.None of the participants reported being native speakers of any language other than English or to have engaged in dictogloss activities in any of their previous language classes.These participants were enrolled in 25 classes: 14 classes (novice-mid level) at University A, and 11 classes (advanced-low) at University B. At the time of the study, University A, a mid-size state-run university, worked under a quarter system.The majority of students enrolled in the novice level series were taking the course to fulfill the College of Arts and Sciences foreign language requirement.In this first year course, students had access to grammar explanations written in English in their textbooks and were provided with explicit grammar instruction while in class, mainly in the L2.They regularly engaged in collaborative activities in the L2 and they were used to receiving input, interacting and producing output.All sections of novice-low, novice-mid and novice-high level classes shared the same syllabus; therefore, the same assignments and exams were administered across sections.On the other hand, at the time of the study University B, a large staterun university, worked under a semester system.The majority of students enrolled in the advanced-low level classes were either majoring or minoring in Spanish.They could have taken a placement test to be automatically admitted to the advanced-low level or have taken the previous required classes in the department.In this third year (advanced-low) course, the students had access to grammar explanations written in English in their textbooks but did not receive explicit grammar instruction as frequently as the other group.They were also familiar with collaborative activities in the L2 and were used to interacting and producing output.All of the sections of advanced-low level classes shared the same syllabus; therefore, the same assignments and exams were administered across sections.Table 1 shows the total distribution of participants.All instructors were new to the dictogloss activity and were therefore trained on how to carry it out.The training period took place during two specifically designed workshops taught by the researcher.Afterwards, all instructors were provided with instructional materials: (a) a handout with directions intended for instructors, (b) a handout with directions intended for students, (c) a PowerPoint containing the warm up activities (preparation stage) and the modeled reconstruction, (d) the dictogloss text to be read to the students, and (e) the questionnaires.

Treatment, data collection and data analysis
All the sections in both institutions engaged in the activity on two occasions.The sessions were conducted on the same day across sections to account for the uniformity needed with regard to materials and content covered.The activity was therefore designed to cover the grammar and vocabulary items included in the syllabus for the given days.Each dictogloss took approximately 35-45 min and the participants were asked to respond to the survey after the second session to allow for a familiarization period due to the complexity of the activity.The task followed the four suggested stages of dictogloss (Wajnryb, 1990), and, as in Swain's (1998) study, the discussion for text reconstruction was "modeled" to foster pushed output.In addition, the texts used for the activity were short, included an average of 6 sentences, were seeded with the target form, and were not so dense with regard to content.
Only participants who were present during both dictogloss sessions were included in the data pool (N = 497).No demographic information was gathered and the responses were anonymous.Frequency was calculated for all responses in order to determine the percentage of respondents selecting a particular answer.

Survey design
Prior to the beginning of the study, the operationalization of the dictogloss and the survey were piloted.One dictogloss session was conducted with two intact classes (one novice-mid and one advanced-low).Afterwards, students completed the survey and were also given a form to provide feedback (both for the task and the survey).Surveys and feedback forms were reviewed in order to identify any problems with wording, to determine if the questionnaire was capturing the information needed to answer the research questions and to estimate how much time the participants needed to complete the questionnaire.
After the pilot, it was determined that the activity was to be implemented following its original format.Additionally, the survey was found too long, and was redesigned to comprise 10 questions.Six of the questions were ordinal and followed a 5-point Likert scale type, 1 was open ended and 3 were multiple choice.Overall, these questions intended to assess aspects of the dictogloss activity and targeted (a) its effectiveness, (b) the perceived skills employed in the activity, (c) the value of metatalk, and (d) the implementation of metatalk.

Results
The responses to the survey items were compared between groups to determine learners' perceptions concerning the effectiveness of the activity as well as learners' engagement in talking and thinking about the language and their appreciation of metatalk.Responses to the 4 survey questions that targeted information related to the research questions are reported here and the remaining data will be reported elsewhere.Significant levels of skewness were noted on nearly every response for each group and are presented in Table, 2 along with the means.Due to a lack of normal distribution, a Mann-Whitney U test was conducted to identify statistical significance between groups.Results indicated that the differences between Group 1 (novice-mid) and Group 2 (advanced-low) were statistically significant (U = 11518.50,z = -11.68,p < .05)regarding the perception of the impact of the activity on their understanding of the targeted form.Concerning Item A in Table 2 ("This activity helped me understand the grammar concept better"), Group 2 valued the dictogloss as a technique that fosters acquisition, whereas Group 1, even though not completely disregarding its value, did not consider this technique as valuable as the participants in Group 2.
Significant effect of group was also found in the responses regarding Item B in Table 2 ("I talked with my partner about how the language works (rules) when we were putting the sentences back together;" U = 17412.00,z = -7.86,p < .05).This indicates that participants in Group 1 did not perceive having talked about the language as much as those in Group 2.Moreover, statistical significance was found on Item C in Table 2 ("I thought about how the language works (rules) when we were discussing with my partner;" U = 18916.00,z = -6.76,p < .05),pointing out that discussion and reflection took place among participants in Group 2 to a greater extent than among those in Group 1.
The same trend existed for Item D (Table 2) concerning the usefulness of metatalk ("Talking and thinking about how the language works helped me understand the concept better;" U = 19562.50,z = -5.90,p < .05),implying that novice-low level learners did not consider metatalk to be an effective practice with regard to their acquisition, while advance-low learners recognized the connection between metatalk and acquisition.To summarize, the dictogloss and its components were overall received differently by the two groups.
Additionally, simple frequency calculations were conducted to establish the percentage of participants selecting one particular answer to the questions that gathered participants' opinions with respect to the effectiveness of the dictogloss, their perceived involvement in metatalk, and the effects that metatalk had on participants' comprehension of a given grammatical form and the overall benefit perceived after completing the task.These responses are summarized respectively in Tables 3 (for Group 1) and 4 (for Group 2).The dictogloss task was considered effective by participants in Group 2 (advanced-low level students), since more than half of them (52%) indicated that they comprehended the concept better after completing the activity.Moreover, in addition to the 35.7% that also agreed with the effectiveness of the activity, only 12% remained neutral and none of them disagreed or strongly disagreed.Conversely, the activity was not so well received by participants in Group 1 (novice-mid level students) since the majority remained neutral (43.4%) and 13.7 % and 2.6% of the participants disagreed or strongly disagreed with the effectiveness respectively.However, 28.1% and 12.5% of the participants stated agreement or strong agreement with the degree of effectiveness, indicating that only half of the participants in Group 1 considered the dictogloss task to be a valuable learning experience, while the majority of the participants in Group 2 considered it effective and none of them indicated disagreement.
The same trend, although not so precise, was revealed with regards to the value of metatalk across Groups 1 and 2. The participants in Group 1 (novice-mid level students) showed a mixed pattern.While the majority (43% agreed, 34.3% strongly agreed) thought that talking and thinking about the language contributes to their understanding of a given concept, 21.7% still remained neutral.On the other hand, participants in Group 2 (advanced-low level students) indicated a distinct preference.Almost 60% strongly agreed with the importance of metatalk in facilitating their understanding of a grammatical concept, 34.6% agreed and only 7.4% remained neutral.
One last consideration should be made concerning the perceived involvement in the different components of the activity, primarily metatalk.While most participants in Group 2 acknowledged having talked about the language with a partner and having thought about the language while conducting the reconstruction stage, participants in Group 1 indicated a less clear perception of discussing and thinking about the language.

Discussion
Concerning the first research question, participants indicated different degrees of agreement about the effectiveness of the task.Advanced-low level learners considered this activity to be greatly effective.On the other hand, novice-mid level students did not consider this activity to be highly conducive to learning.This could be attributed to factors such as the length and complexity of the task, which are elements that could interfere with its applicability at novice levels.This collaborative task entails interaction and output production, two procedures that can promote noticing related to meaning or form; in addition, metatalk is expected to occur especially during the reconstruction period, particularly if the reconstruction was previously modeled.Metatalk is thought to raise awareness and promote noticing (Swain, 1998), which will consequently have a positive impact on the development of learners' interlanguage.Consistent with García Mayo (2002), the results of the present study indicate that the auditory component of the dictogloss might have posited a greater challenge for novice-mid learners than to higher-level ones.It could be speculated that the participants in Group 1 were more concerned about employing their listening abilities and preoccupied with determining form-meaning connections.
It seems that the reconstruction stage was perceived by the novice-low students as a mere regurgitation of the words and phrases they were able to capture during the listening rather than an opportunity for discussing formrelated instances in order to reconstruct the sentences.Moreover, directing their attention towards other aspects of the task such as the recognition of lexical items or the accuracy of the notes taken could have limited students' capacity to attend to target forms and to engage in metatalk.
Participants in Group 2 possessed a level of competence that allowed them to comprehend the passage heard without much difficulty and consequently feel more confident about the accuracy of the notes taken, which gave them more time for focusing on form and discussing specific linguistic aspects in order to reconstruct the text.Therefore, novice-mid level participants (Group 1) could have been confused as to why a task that appeared to be a mere listening activity also required them to engage in metatalk.On the other hand, advanced-low level participants (Group 2) could have been less concerned about the adjacent aspects of the task and were cognitively ready to conduct the reconstruction and engage in metatalk, focusing on form rather than primarily or exclusively on meaning.Advanced level learners are expected to be cognitively ready to notice and produce certain complex forms, whereas beginner or intermediate-level learners might still be at a stage in which they intend to grasp meaning and might experience difficulty when expected to focus on form (Leeser, 2004).
It appears that participants in Group 1 (novice-mid) were more prone to focus their attention on comprehending lexical items in order to enable their construction of meaning, therefore restricting their chances of noticing forms to facilitate the reconstruction of the text.Given that attention is limited, due to economy principles, it is also selective and must be, therefore, strategically allocated.Schmidt (2001) indicated that "the allocation of attention is the pivotal point at which learner-internal factors (including aptitude, motivation, current L2 knowledge, and processing ability) and learner-external factors (including the complexity and distributional characteristics of input, discoursal and interactional context, instructional treatment, and task characteristics) come together" (pp.12-13).For this reason, advanced-low level participants (Group 2) could have been able to allocate their attention to a wider spectrum (due to higher processing ability, L2 knowledge, aptitude, and motivation).
The second research question aimed to investigate whether participants were able to recognize the acts of talking and thinking about the language as inherent components of the dictogloss activity.Results showed a discrepancy between the two groups indicating that those in Group 1 conducted the activity without realizing that the objective of reconstructing the text was to discuss how the language works.Those in Group 2, on the other hand, were well aware of having talked and thought about the language.This indicates that the ability to understand the central point of this activity and to recognize that reflection and metatalk were taking place could be related to students' level of L2 competence, familiarity with collaborative tasks and cognitive readiness.Interestingly enough, contrary to the results shown in Swain (1998), findings of the present study do not seem to indicate a connection between modeling the reconstruction and metatalk, and the participants' ability to recognize their involvement in metatalk.In the present study, participants in both groups benefited from a modeled reconstruction and were instructed to pay special attention to certain features, and yet they were able to recognize their engagement in metatalk at different rates, due to their proficiency level.
Lastly, the third research question was established to investigate whether participants valued metatalk.The effectiveness of metatalk has been demon-strated through previous research, and even though this study did not measure the number of LREs produced or participants' ability to engage in metatalk, results are in line with those of Fortune (2005), whose advanced-level participants indicated a higher level of readiness to engage in metatalk when compared to intermediate-level participants.Our advanced-low participants (Group 2) valued metatalk significantly more than the novice-mid participants (Group 1).Gutiérrez (2008) pointed out that the ability to reflect about language is related to the ability to engage in metalinguistic activity.Therefore, it can be concluded that the ability to engage in metatalk, which is related to the ability to reflect about language, can also be linked to learners' ability to recognize this as a practice conducive to their language acquisition.Moreover, in spite of the fact that students in both groups valued metatalk as a practice that impacts positively their acquisition process, the majority of the participants in Group 2 only recognized having talked (63.3%) and having thought (62%) about the language while engaging in the activity.By contrast, a mere third of the participants in Group 1 identified talking (33.2%) and thinking (35.9%) about the language as part of the dictogloss activity.What could explain this is that participants who were not cognitively ready to embark on a lengthy and complex task were not able to recognize the fact that discussing the language was a focal point and were therefore less able to perceive metatalk as a valuable learning experience.
One last consideration should be made concerning participants' ability to recognize their involvement in metatalk.Gutiérrez (2008) has claimed that metalinguistic activity can be produced in other ways than explicitly (through recitation of rules or discussion of said rules), indicating that when not verbalizing their noticing, it is not necessarily true that learners are not able to partake in metalinguistic activity due to lack of knowledge.We can, therefore, speculate that, especially in the case of novice-mid level learners, the lack of verbalization might not necessarily indicate lack of knowledge, or lack of metalinguistic ability.This is attributed to the fact that participants in Group 1 recognized thinking about the language more than discussing it with their partners, whereas the level of recollection about their involvement in both aspects (thinking about the language, talking about the language with their partners) was very similar with respect to both actions amongst participants in Group 2. Therefore, novice-mid level learners might possess the capacity to engage in metalinguistic reflection, which could indicate metalinguistic ability, and yet either decide not to, or lack the ability to engage in discussions about the language, especially when required to do so using their L2.As previously mentioned, participants in the present study were encouraged to avoid the use of L1, but this could not be controlled as strongly with the novice-mid level participants, who did not have the appropriate competence to conduct metatalk in the L2.

Implications, caveats and future research
Results indicate that advanced-low level learners considered dictogloss as more valuable in comparison to lower-level learners; however, its applicability across levels can be fostered by making some adjustments.For example, during the first stage it is important to discuss the topic of the text and to introduce new vocabulary.The topic of the text should be appealing and familiar to the students since it will facilitate their interaction before engaging in the task and their commitment to the task in general.Also during Stage 1, the reconstruction process can be further explained by indicating that this is not a regular dictation and sentences should not be copied verbatim, as well as clarifying that this is not a listening comprehension activity and that students should not concern themselves with exact reconstruction.During the dictation and reconstruction stages no feedback should be provided, except for the clarification of instructions.The last stage is critical to the successful completion of the activity and, according to Wajnryb (1990), Kowal and Swain (1994) and Swain (1998), it is as important and valuable as the reconstruction-interaction stage.Since the provision of feedback is very relevant during this stage, it is recommended to first review the sentences each group produced.If the time permits, all sentences may be revised; if not, one sentence per group should be sufficient.Additionally, given that metatalk can be conducted not only related to the sentence level, LREs might focus on features such as lexicon or discourse because the discussion could incorporate these aspects as well.
Dictogloss has been implemented in various ways in different studies; therefore, the interpretation and generalization of the results could be problematic if the premises originally established for the task were not followed.Consequently, future research should measure the effectiveness of dictogloss as long as all four stages are completed and any modification made to the implementation is carefully explained.
Fostering familiarization amongst learners before carrying out this activity could contribute to a richer collaboration and more fruitful metatalk.Participants' level of competence in the L2 as well as the activity's complexity leading to cognitive overload should be considered when implementing this task with novice learners.
The results yielded by this investigation should be considered in light of the limitations it presents.Firstly, the degree of receptiveness the participants displayed towards this activity can be related to their overall attitude towards the class, the instructor, and L2 learning in general.Collecting data at two different universities might have also impacted the results.The present investigation concerning students' attitudes towards dictogloss across levels contrib-utes to determining its effectiveness and instructional value.However, since the instructors' impressions were not investigated, future research could examine that as well as the potential connection between instructors' attitudes and students' attitudes.Additionally, more research is needed concerning learners' ability to engage in metalinguistic reflection and their ability (or lack thereof) to engage in metatalk and the possible causes of such connections.Lastly, this study did not examine the interrelation between learner attitudes and degrees of receptiveness or the amount and types of LREs and L2 acquisition, which could be investigated in future research.
Overall, although several studies that have evaluated dictogloss have focused on the amount and type of LREs produced during the reconstruction stage, fewer studies investigated the effects this FonF technique has on the acquisition of a given form and interlanguage development.This could be further investigated taking into account not only the type of metatalk but also any possible correlation between types and amounts of LREs produced, the overall effectiveness of the activity, the level of receptiveness displayed by learners, the correlation between focus on a specific form through collaboration and metatalk and acquisition of that form, and the development of L2 fluency and accuracy.

Conclusion
Findings from this study indicate that a collaborative task such as dictogloss is generally well received and valued by students, more so by those who display higher levels of L2 competence and who are cognitively ready.Even when the metatalk and the reconstruction are modeled for students, the different components (the listening aspect, the interaction, the reconstruction of the text, the metatalk conducted while reconstructing, the form-meaning connections, etc.) can posit a considerable amount of challenge for learners of lowerproficiency levels.However, despite facing more challenges when engaging in the task, novice-mid learners also valued dictogloss as a fruitful learning experience.Data presented here is therefore in line with the studies favoring the implementation of dictogloss as an effective FonF technique.

Table 1
Participants and institutions

Table 2
Descriptive statistics

Table 3
Frequencies and percentages for novice-mid level participants (Group 1)

Table 4
Frequencies and percentages for advanced-low level participants (Group 2)