A cross-sectional study on task type and negotiation of meaning in CLIL child-child interaction

Different task types have been hypothesised to affect the type and amount of Negotiation of Meaning (NoM) generated in learner-learner interactions. However, studies specifically addressing the impact of the task variable on the NoM in child-child interactions in foreign language contexts are virtually non-exixtent. This study analyses the amount and type of NoM operationalised as conversational adjustments (CAs) present in the interactions of primary education L1-Spanish young learners (YLs) of English on two different tasks. Participants were 40 eight-year-old children enrolled in a partial immersion Content and Language Integrated Learning (CLIL) program. Ten pairs took part in a two-way pictureplacement jigsaw, while the other ten participated in a one-way picture-placement storybased task. The results indicate that the impact of task type on the amount and type of CAs produced by participants is more far-reaching than expected, revealing a significantly greater amount of CAs in the one-way task. This finding locates this variable at the very core of the list of factors directly impinging on NoM.


IntroductIon
There is ample evidence that tasks are considered very efficent tools for second and foreign language acquisition (García Mayo, 2007;Pica, 2013;Pica et al., 2006;Van den Branden et al., 2009). Accordingly, methodologies such as task-based language teaching, are becoming commonplace in language classrooms. Tasks provide a meaningful context in which learners can use and test their knowledge of the target language (TL), and foster mental processes essential to language learning. From an interactionist perspective, tasks are particularly beneficial when they promote interaction among participants, even more so when negotiation of meaning (NoM) takes place (Long, 1996).
A number of studies have addressed the effects of different task types and task conditions on task performance and subsequent learning, showing that task type constitutes a crucial factor (Bygate, 2001;Philp et al., 2006;Robinson, 2011)with a view to better understanding the structure of tasks, their impact on students, and their use by teachers. This edited volume starts with an introduction to the background and key issues in the topic area. Each section begins with a succinct introduction, and the volume concludes with an afterward relating the theme of the volume to issues in curriculum development. The book is divided into 10 chapters: \"Effects of Task Repetition on the Structure and Control of Oral Language\" (Martin Bygate. However, research on the connection of task type and interaction is scarce and barely exists if we focus on young learners (YLs). Languages are being taught to learners of different ages, and young language learners constitute a population which increases rapidly the world over (García Mayo, 2018;Pinter, 2017). Still, to the authors' best knowledge, very few publications are available in the literature that discuss how task type influences YLs' performance (Azkarai & Imaz Agirre, 2016;Oliver, 2002;Oliver & Azkarai, 2019).
It is essential to establish how different tasks promote different aspects of YLs' language learning process. From the interactionist framework, the present study intends to shed light on this matter by examining the episodes of NoM that take places among YLs of English as a Foreign Language (EFL) while completing two different task types, namely a jigsaw and an information gap task.

Tasks in Interaction-based Studies
Tasks have been proven to be beneficial for second language learning and have extensively been used in interaction-based research (García Mayo, 2007;Van den Branden et al., 2009). During interaction, learners resort to a variety of strategies to negotiate for meaning. NoM is possible and beneficial for all learners, children and adult, for it is said to "connect input, internal learner capacities, and output in productive ways" (Long, 1996, pp. 451-452). Canonical NoM strategies include conversational adjustments (CAs) and different types of repetition. CAs are used to increase comprehensibility and conventially consist of the following three types: clarification requests, confirmation checks and comprehension checks.
Tasks have been categorised according to the interactional criteria, the direction of the information flow and the outcome (Pica et al., 1993). Tasks that require interaction among learners and have a single, convergent outcome are considered as the most effective because they trigger comparatively more opportunities for NoM than other, less controlled types of interaction such as those elicited in decision-making and opinion exchange tasks (Pica et al., 1993(Pica et al., , 2006. In this respect, a qualitative NoM-based study by Nakahama et al. (2001) on adult Japanese EFL intermediate-level learners interacting with native speakers of English compared instances of NoM in conversational tasks, that is, non-structured conversation, and a spot-the-difference task (i.e., a task with a single, convergent outcome). Among other findings, their study indicated that the conversational task did offer fewer instances of NoM than the spot-the-difference task. However, these researchers highlight the fact that such NoM revolved around lexical items solely, to the extent of it becoming "mechanical" or, as the authors put it, too "strictly informational focussed" (Nakahama et al., 2001, p. 388). Similarly, Underhill (1987) also warns that information gap tasks -equally requiring a single, convergent goal and outcome-often involve the risk of focussing on general problem-solving ability more than language fluency (Underhill, 1987). Nevertheless, all those studies were carried out with adult populations, and their findings should not be transferred to children populations by default (Mackey et al., 2003).
In fact, the participants in the present study have two characteristics that make them significantly different to those in the studies mentioned above: they are children, and they are beginner learners of the TL. Some scholars have argued that children are considered to rely on adults, or more proficient speakers, to manage conversations for them (Lázaro-Ibarrola & Azpilicueta-Martínez, 2019;Scarcella & Higa, 1981). It could be hypothesised, then, that, in the absence of adults or more proficient speakers, clear, goal-oriented tasks might scaffold child-child oral collaboration more than less-structured conversation. What is more, Azpilicueta-Martínez (2017) reported significantly higher NoM rates in child-child interaction than in child-adult interaction when performing the same task.
Regarding the flow of information (i.e., one-or two-way), the latter appears to be more balanced in terms of learning opportunities for participants, since all of them hold part of the information needed to successfully perform the task. One-way tasks, by contrast, foster a unidirectional type of flow because the learner in need of the information would have to negotiate for information more than their partner (Gass & Varonis, 1985). Consequently, tasks have become a fixture in recent interaction-based studies (Azkarai & García Mayo, 2017;García Mayo & Lázaro-Ibarrola, 2015;Hidalgo, 2019;Pinter, 2006Pinter, , 2007. However, a study by Galaczi (2014) on the interactional competence of learners revealed that, the lower the level of command of the TL, the higher the difficulty in maintaining a balanced interaction between participants. She postulated that this difficulty in keeping both the speaker and listener role active simultaneously could be due to the high cognitive demands on that type of interaction, in which low-level learners would struggle to both decode their partner's speech and compose their own contributions, given their limited free working memory (Galaczi, 2014).
To date, only Azkarai and Imaz Agirre (2016) have specifically addressed the impact of these two types of tasks on NoM rates in child-child interaction 1 . They did so via a one-way guessing game and a two-way picture placement task performed by fourth and sixth-grade students, each of which age groups included, in turn, mainstream EFL and Content and Language Integrated Learning (CLIL) subgroups. Their findings revealed significant differences between the mainstream learners in both age groups, with the one-way task yielding significantly higher rates in different CAs. The two-way task, on the other hand, generated significantly higher CA rates among the 6 th grade CLIL students, i.e. the more mature and linguistically competent students. Interestingly, their 4 th grade CLIL learners were the only ones not displaying any significant differences between tasks, although, in this age group, the one-way task did trigger higher rates in clarification requests, confirmation checks, and overall CA rates.
In addition to the above, there is research supporting the notion that complexity levels in tasks of the same type also affect NoM. Complex tasks are believed to trigger comparatively more NoM than simpler tasks (Robinson, 2001(Robinson, , 2011and for motivating sequencing decisions in task-based syllabuses. Results of a study of the relationship between task complexity, difficulty, and production show that increasing the cognitive complexity of a direction-giving map task significantly affects speaker-information-giver production (more lexical variety on a complex version and greater fluency on a simple version, yet such direct correlation has been called into question by several researchers. Nuevo (2006) focussed on CAs, among other interactional features, and found out that different levels in task complexity seem to promote different CAs. In her study, simpler tasks were found to lead to higher uptake rates in comprehension checks and other-repetitions than the more complex versions of the same tasks. Similarly, Gilabert et al. (2009) suggest that, although more complex versions of a given task might yield comparatively higher CA rates, the scope of such differences seems to hinge on task type (Gilabert et al., 2009).
All in all, the interaction between task typology and NoM appears to be an extremely complex issue on which research findings so far seem far from conclusive.

Research questions
In order to further understand the role of task type on YLs' NoM, the present study addresses the following research questions: 1. Do one-way and two-way tasks generate similar amounts and types of CAs? 2. How do the learners in the present study compare with the YLs in previous research (i.e. Azkarai & Imaz Agirre (2016))?
On the basis of the literature reviewed, we expect to find NoM among the participants in the two groups, along with differences in the amount and type of the NoM strategies each task triggers.
We anticipate that both tasks will foster NoM since these tasks types require interaction among the participants in order to achieve a common goal (Pica et al., 1993(Pica et al., , 2006, but no clear prediction can be established. Based on adult studies, the higher complexity of the story-based task might elicit more NoM (Robinson, 2001(Robinson, , 2011. However, based on the study carried out with YLs, there might be no differences between the two tasks (Azkarai & Imaz Agirre, 2016).

Participants
Forty children at beginner levels of competence of the TL who were enrolled in a partial immersion bilingual program participated in the study, and were divided into two groups. Each group comprised 20 children, 11 girls and 9 boys in one of the groups, and 13 girls and 7 boys in the other. Each group performed only one of the two different tasks explained in section 3.3 (The tasks). All participants were studying their Primary Year 3 course, and their mean age was 8.5.
At the beginning of the study the children were told that they were going to take part in a game in English, and that the tasks themselves were not meant to be marked in any way. Parents were informed that their children's performances would remain anonymous and limited for research purposes only. Due permission was granted by parents and the school.
The proficiency of spoken English of the children was based on the schools' internal assessment records, as well as on their performance in diagnostic testing carried out by external examiners from the local administration, which placed the participants' oral proficiency level at pre-A1/A1 of the Common European Framework of Reference for Language (CEFR).
The participants' main characteristics as regards age and proficiency are illustrated in Table 1 below:

Instructional setting
All subjects had had an average 10-hour weekly exposure to the TL (English) at school for five and a half years. Approximately 50% of such exposure had included explicit EFL instruction, while the remaining 50% had comprised subjects being taught through the medium of English, i.e. CLIL. They had done so at two different state Primary Schools in the north of Spain, both of which shared the same bilingual program from the Ministerio de Educación, Cultura y Deporte (MECD)-British Council 2 . All children shared Spanish (or Spanish plus an additional language in the case of bilinguals) as their L1, and access to English-speaking interaction outside their classes was limited.

The tasks
The two tasks in the present study share a common structure: they both require interaction among participants and have a single outcome, in order to promote an exchange of information and the generation of NoM. Neither task had a time limit set for the learners. However, Task A is a jigsaw (two-way), whereas Task B is a story-based information-gap task (one-way). Close attention was paid to designing tasks which could not be successfully performed without the effective exchange of linguistic information.
In order to achieve ecological validity, the tasks were designed by the authors in collaboration with the participants' teachers. The tasks were designed with the aim of implementing them within real classroom conditions. Task design was based on others used in previous studies: a jigsaw and an information-gap task were selected because, as already stated, these task types are considered the most appropriate to foster interaction (Butler & Zeng, 2014;Oliver, 2009;Pica, 2013;Pica et al., 1993).
Thanks to the opportunities for interaction, these task types offer conditions for learners working in pairs to negotiate for meaning. Since mutual understanding is needed, speakers sometimes need to clarify or explain the meaning they intend to convey (thus producing modified output and comprehensible input) and offering feedback to their interlocutors in response to their output. This way, these tasks provide learners with plenty of opportunities to focus on meaning, function and form (Pica et al., 2006).
Likewise, the children were not given planning time in order to boost talk and more constructive on-task behaviour (Philp et al., 2006).

Task A
Task A, the jigsaw, was mainly designed to shed light on YLs' interactional patterns and, to a lesser extent, it intended to establish a context for description at a basic level. With regard to the characteristic real-world target of pedagogic tasks, the posters used portray real-life scenes, showing places well-known to the participants in this study (i.e., a classroom and a playground), thus allowing the learners to experience and use authentic and meaningful language.
To perform this task, two identical posters (one for each learner) and two identical sets of 6 photos of children (these pictures will be referred to as a, b, c, d, e, f) were designed. Student A in the pair had her poster with pictures a and b already placed on it, and pictures c, d, e and f outside the poster, while student B had the other poster with pictures c and d on it and pictures a, b, e and f outside the poster. Thus, participants had to find out which two pictures were the ones on their partner's poster. The set of pictures included two distractors that shared some features with the target ones in order to further promote interaction among the participants (See Appendix). Each pair of students sat at two tables separated by a screen so that they could not see each other, and were forced to rely exclusively on oral English. The goal of the task was that the participants, without seeing each other or each other's materials, interacted in English in order to complete their posters so that both ended up having the same pictures in the same places on their respective posters. In order to do so, the participants had to use the TL to ask questions as well as provide information to their partners.

Task B
Task B consisted of two stages, each of which revolved around a single 5-picture story that one child narrated to their partner, who was given the instruction to build the same story by selecting the right pictures from a set of 8 jumbled-up cards comprising the 5 correct items plus 3 distractors. They then had to ask the narrator questions in order to find out the right pictures as well as their position within the story. Consequently, Task B is one-way because only one of the members in each pair holds all the information required to complete the task.
Task B outlined two well-defined roles: a narrator and an active story builder; hence the need to perform the task twice (each time with a different story and set of pictures), in order for all participants to perform each role once. Otherwise it could be hypothesized that students in the story builder role would have to negotiate for meaning more than the information holder (Gass & Varonis, 1985), as opposed to Task A, in which both partners shared identical roles. In addition, the design in Task B not only forced story builders to spot the right picture, but also to place each of them chronologically within a storyline. Correspondingly, narrators had to describe pictures within a plot of sequential events. Differences between some of the images and their distractors were intentionally not clear (Duff et al., 2008) in order to generate misunderstandings and trigger negotiation. Likewise, to minimize predictability within the storylines, the pictures in them were occasionally presented illogically in terms of the chronological order or coherence within the story (see Appendix), so that participants could not 'anticipate' which picture in the story was more likely to be a distractor without the need to interact. Task B constitutes, thus, a hybrid task comprising an information-gap task embedded within a story-telling activity. This was done with the purpose of promoting interaction (Pica et al., 1993). Table 2 summarises the main characteristics of the two tasks used in the study. Without seeing each other or each other's materials, the participants had to interact in order the story builder to have the same story as the narrator, with the same pictures, in the same order.

Information flow
Two-way One-way

Exchange of information Required
Outcome Closed

Practice
In order to maximize participant-friendliness and to facilitate an instinctive understanding of the tasks, participants performed similarly-structured tasks (although based on different storylines and lexical elements) one week prior to data collection. This was done by the teachers during their actual class time (Task A), and by one of the researchers (Task B). However, the data for the present work, as in many other research studies (e.g. Bagheri et al., 2012;Sample & Michel, 2014), were collected in a controlled environment with the researcher and the pair of learners outside the classroom. This setting guarantees a uniform implementation of the tasks across the different participants, even though we are aware of the concerns about the validity of this type of data.

Data analysis and codification
Due to the small sample sizes, the non-parametric Mann-Whitney U-test was used. Significance level was fixed at p= 0.05. All quantitative analyses were conducted using SPSS Version 24.
Students' performances were recorded (total recorded time 4 hours 52 minutes) and transcribed by the two researchers in the present study. The participants' production was then coded according to Oliver's classification of CAs (Oliver, 1998), i.e., clarification requests, confirmation checks, comprehension checks, to which we also added the inclusion of acknowledgements of understanding (Lázaro-Ibarrola & Hidalgo, 2017a). CAs have been illustrated with examples from our own database: i) Clarification requests: "Any expression [...] designed to elicit clarification of the interlocutor's preceding utterance(s)" (Long, 1983, p. 137 In Example 1, Student A seems to fail to understand what student B has just said and requests clarification. That clarification request triggers Child B to repeat her previous utterance, which leads, in turn, to the provision of the information by Child A, confirmed by an acknowledgement of understanding (explained below) by Child B (turn 5). As can be seen in this example, NoM does not always lead to output modification. After Child A's clarification request, Child B answers by repeating her exact previous words. Other instances of NoM are more successful in terms of modification of the learners' initial output. The NoM triggered by the clarification request in example 2 forces Child B to provide the information requested by modifying his output, initially unintelligible to Child A.
ii) Confirmation checks: "Any expressions […] immediately following an utterance by the interlocutor which are designed to elicit confirmation that the utterance has been correctly heard or understood by the speaker" (Long, 1983, p. 137). This may be noted in the following example (3). In example 3 Child B seeks confirmation that she has fully understood the information, namely that the two children in the picture are girls 3 . This CA has triggered Child A to modify his output by specifying it is indeed one girl and one boy.
iii) Comprehension checks: "Attempts to anticipate and prevent a breakdown in communication" (Long, 1983, p. 136), i.e., they constitute the only CA which is performed by the speaker in order to guarantee that their interlocutor understands, as opposed to clarification requests and confirmation checks, which are aimed at achieving the speakers' own comprehension: In example 4 Child A asks whether he should move on and describe the following picture. Comprehension check rates among primary school children have been traditionally scant, a fact which has been hypothesised to be related to the egocentricity of this age group, when children seem to be more interested in creating their own meaning than in checking whether their partner understands them (Oliver, 1998) (but see Lázaro-Ibarrola & Hidalgo, (2017a) and Hidalgo (2019)). iv) Acknowledgements of understanding: "What takes place when a supportive listener in a pair offers verbal signs of comprehension or provides audible support to the speaker." (Ducasse, 2008, p. 94), or, as Galaczi (2014 puts it, "evidence of the ability of the listener to monitor what is being said" (Galaczi, 2014, p. 567 [Acknowledgement] 7. Child A: Do you have a girl with a t-shirt gray?
Example 5 illustrates how the participants resorted to words like "ok" to indicate their partners that they have understood what was previously said, and that they could move on to a next step within the task.
The analysis of each strategy was carried out considering the total number of strategies the participants used divided by the total number of utterances. An independent rater coded 25% of the data set. Inter-rater reliability was calculated using simple percentage agreement, which resulted in 93.5%. All remaining discrepancies were solved on a case-by-case basis.

reSuLtS
The first research question in the present study intends to ascertain the extent to which the task type factor may affect the amount and type of CAs generated in the interaction of young EFL learners.
The following table (3) compares the results for CAs identified in Tasks A and B. Raw numbers, percentages and the results from the statistical analyses are provided in the table. Results reveal considerable differences in the number of strategies identified in each task. The greatest differences are located in the number of clarification requests and acknowledgements, as illustrated in Table 3. Consistent with previous studies, barely any comprehension checks were identified in our dataset. The results display statistically significant differences between tasks in most CAs, namely clarification requests (U= 106, p= .006), comprehension checks (U= 140, p= .009) and acknowledgements (U= 65.5, p< .001), with Task B (one-way) generating more instances of all three. As it was expected, overall CAs use was statistically significantly more frequent in the data obtained in Task B (U= 55, p< .001). Acknowledgements are, clearly, the single CA with the widest gap between groups. Confirmation checks in the two-way task (Task A) constitute the only CA showcasing higher rates than the one-way task, yet such difference was not statistically significant.
In addition, in the one-way task, two clearly prominent strategies have been identified: acknowledgements (N= 66) and clarification requests (N= 29), whereas in the two-way task the differences between the individual NoM strategies identified are not so well defined (See Table 3) as the numbers are low in all categories.
The second research question intended to compare our results with the CLIL 4 th graders in Azkarai and Imaz Agirre (2016). Results are shown in Table 4. Here we have only considered the CAs included in Azkarai and Imaz Agirre (2016), i.e., clarification requests, confirmation checks and comprehension checks. Due to the fact that individual rates in Azkarai and Imaz Agirre (2016) were unknown to us, statistical analyses could not be carried out and results are limited to raw numbers and percentages. In the present paper, the YLs performing the one-way task have produced more CAs (to a statistically significant degree) than the participants in the two-way task. This finding contrasts with Azkarai and Imaz Agirre (2016), who reported that the differences between the performance of the CLIL 4 th graders did not reach statistical significance. Nevertheless, it is also possible to infer commonalities between them. Clarification requests, and the average of all three CAs, displayed higher rates in the one-way tasks in both studies, and comprehension checks were nearly non-existent in the four groups.

dIScuSSIon
The present study attempted to shed light on the degree to which variations in the task factor affect YLs' NoM, and to offer further evidence to the findings presented in Azkarai and Imaz Agirre (2016), a study which compared CA rates in one-way and two-way tasks.
The results in our study pinpointed statistically significant differences between the two tasks types, with the one-way task leading to higher NoM rates in all CAs with the exception of confirmation checks. Results in the groups closer in age and instructional setting to the learners in the present study in Azkarai and Imaz Agirre (2016) revealed no statistically significant differences between tasks, yet overall CA rates, as well as the two most frequent NoM strategies (i.e., clarification requests and confirmation checks), were higher in the one-way task (a guessing game). Consequently, it seems that one-way tasks could constitute an efficient means to generate NoM with young EFL learners. These results are consistent with Azkarai and Imaz Agirre (2016), who suggest that YLs in this particular context (CLIL in Spain) seem to engage in more NoM episodes when carrying out one-way tasks. It seems that, when working with this task type, learners try to fully understand their partners' message, perhaps because that is their only chore/role. However, in two-way tasks, the participants act both as information-requesters and information-holders simultaneously (Pica et al., 1993). This reason might explain why not so much NoM takes place in two-way tasks: even though a priori this task type may seem to be more balanced, it may also be more demanding for EFL YLs. However, the degree and type of NoM produced by learners might also be affected by differences in task type (Gilabert et al., 2009), i.e., whether it is a guessing game or a picture-based storytelling game (Task B), or complexity (Robinson, 2001(Robinson, , 2011and for motivating sequencing decisions in task-based syllabuses. Results of a study of the relationship between task complexity, difficulty, and production show that increasing the cognitive complexity of a direction-giving map task significantly affects speaker-information-giver production (more lexical variety on a complex version and greater fluency on a simple version, since Task B appeared to be comparatively more complex. Also in line with Azkarai and Imaz Agirre's (2016) findings, our results suggest that, whereas in one-way tasks YLs seem to resort to acknowledgements and clarification requests when they encounter a communication difficulty, in two-way tasks the tendencies are not well defined and YLs, despite negotiating less, seem to rely on a wider array of NoM strategies.
Concurring with previous research on EFL child-child interaction (e.g. Azkarai & Imaz Agirre, 2016;Lázaro-Ibarrola & Azpilicueta-Martínez, 2015;Lázaro-Ibarrola & Hidalgo, 2017a,b;Pinter, 2006)which have been claimed to lead to second language learning. However, research on child interaction in foreign language settings is scarce, specifically research on a new prevalent methodology in Europe, content and language integrated learning (CLIL comprehension checks were the lowest CA, which might support Oliver's claim that children's low rates of this CA type relate to their egocentricity and limited ability to focus on their interlocutors' needs (Oliver, 1998(Oliver, , 2009). However, our participants also negotiated to let their interlocutor know that the message had been understood by resorting to acknowlegments of understanding. These instances of peer assistance provide evidence of YLs concentrating not only on their own needs but also on their partner's (Hidalgo, 2019;Lázaro-Ibarrola & Hidalgo, 2017a;Pinter, 2007).
Additionally, the high acknowledgement rates in the one-way task in our study suggest that such CA may render the use of comprehension checks redundant. In other words, a confirmation of understanding by the listener via an acknowledgement would eliminate the need for the speaker to seek such confirmation by means of comprehension checks (Lázaro-Ibarrola & Hidalgo, 2017a). Results suggest that this could become more apparent in one-way tasks in which one of the learners is holding the information and their partner is clearly seeking it and is forced to negotiate more. It would be of high interest to find out whether this hypothesis is confirmed in subsequent interaction-based studies. Likewise, it is important to highlight that acknowledgements of understanding might be more subtle than the more canonical forms of confirmation of comprehension would suggest. A non-canonical expression of comprehension like "ok", followed by change in the subject being discussed by the participants would implicitly mean "I hereby understand what you just said", an explicit phrase which might not seem a feasible utterance with learners this stage. Therefore, careful attention must be paid by researchers in order to accurately code a wider variety of instances of CAs.
Of interest is also the fact that NoM in our study, as in Nakahama et al.'s (2001), tended to focus on lexical items predominantly, and there was a reminiscent mechanical side to the interactional patterns, as the following excerpt from Task A illustrates: The fact that differences between pictures hinged on lexical items and prepositions of place might have served both as a scaffold, guiding the language elicited more closely than an open conversation would, but also as a constraint to the type of discourse and NoM generated.

concLuSIon, LImItatIonS and LIneS for further enquIry
In conclusion, the present study constitutes an attempt to better understand the impact of task typology on the CAs of age-and level-matched EFL children interacting orally in a CLIL setting. Our findings indicate that, as expected, task typology plays a pivotal role in the NoM in EFL child-child interaction. Results have revealed that, at this age and level, one-way tasks trigger more NoM than two-way tasks, probably because learners can fully concentrate on their role, i.e. information-holder or information-requester, thus freeing more resources to focus on meaning and form. Likewise, the comparatively higher complexity level of Task B might have also contributed in this respect, yet further studies are needed in order to support this first hypothesis and shed more light on these issues of unquestionable research interest.
One of the limitations of the present study lies in the fact that no placement tests were used in order to measure the participants' level before data collection. As explained above, an oral-based test would have been a more desirable option, yet task similarities with oral proficiency testing formats at this level and age (pre-A1/A1) might have impinged on the subjects' production, and were discarded. Both researchers agreed on the schools' internal assessment plus external local administration results in order to reach a compromise solution as to the participants' selection and pairing.
Likewise, a certain level of familiarity with the procedural aspects of the task might constitute another limitation to this work. While participants remained fully unfamiliar with the content, tasks with identical procedures (i.e., jigsaw and story-based information-gap) were rehearsed one week prior to data collection with the purpose of getting students acquainted with the particular mechanics of each task. Also, it is worth mentioning that, while students did know their partners and researcher, the fact that the tasks themselves were not carried out in their ordinary classroom might affect the ecological validity of our study.
As a whole, the young EFL learners in the present study managed to interact with same level-and-age peers, and were able to negotiate for meaning in order to understand their interlocutors' message and complete the tasks successfully, without external intervention. This study, consequently, contributes to show that teachers might benefit from interaction-based activities in student-student interaction, even at this low level and age.
Our results also appear to confirm the notion that more complex tasks might promote comparatively higher NoM rates than simpler ones, yet also seem to indicate that, in the case of this population, one-way tasks might make up for children's not fully developed cognitive ability to hold the speaker and listener role active simultaneously, as demanded by two-way tasks.
Given the relevance of interaction relative to EFL learning, findings in the present study support the idea that task features constitute a central element affecting the amount and type of NoM. Although more research is needed, results suggest that age-and-level suited complex one-way tasks might constitute an ideal tool for the promotion of NoM of children at this age and level, and should be seriously considered by CLIL and EFL teachers and educators.

acknowLedgementS
This study was supported by grant FFI2012-32212 (Spanish Ministry of Economy and Competitiveness), which is hereby gratefully acknowledged. We would also like to thank the schools, the teachers and all the children for their willingness and generosity to participate in this project.

THE TASKS (author's elaboration)
Task A: