"I'm just gonna put 'la' in there, everywhere": researching cognitive and metacognitive strategy use in tasks focusing on Italian object pronouns.

Strambi, Antonella ; Kennedy, Claire ; Dekker, Wendy 等

Abstract: This pilot study aims to contribute to scholarship on Language Learning Strategies (LLS) for learning elements of L2 grammar. We conducted an exploratory analysis of cognitive and metacognitive strategies used by learners in tasks focusing on Italian Object Pronouns (OP). Think-aloud data were collected from four intermediate-level students of Italian at an Australian university. Their verbal reports were transcribed and coded using a taxonomy based initially on that of Chamot et al. (1988a), but modified during the process. Based on our results, we suggest that future studies investigate relationships between strategy frequency, time on task and task scores, while controlling for task difficulty. Qualitative differences in the use of cognitive strategies such as Deduction, Translation and Inferencing could also be explored in relation to task achievement. Finally, exploring the influence of personality traits and gender differences on the use of Self- evaluation and Elaboration could yield interesting results.

Keywords: Object pronouns, cognitive strategies, metacognitive strategies, grammar, teaching and learning.

Introduction

That learning Italian object pronouns (OPs) is very demanding for Anglophones has been recognised by various authors (Zorzi; Pona amongst others). Reporting on her experience as a learner of Italian in the humorous book La Bella Lingua, Dianne Hales lamented: "With even greater effort I struggled to corral its impish pronouns, which flit from the front to the back of sentences, disappear entirely, or latch on to verbs like fleas to a cat's ear."

Despite these recognised difficulties, research on the process of acquiring and using Italian OPs has been relatively scarce. Some empirical studies have reported on natural acquisition orders (e.g. Berretta), and the types of errors that learners of Italian as a second language (L2) tend to make when using OPs (e.g. Pona; Costi; Zorzi). However, little is known about the strategies that learners employ in order to make their work more effective and efficient in this area of Italian grammar, especially in instructed environments.

Research into the acquisition of clitic pronouns in other Romance languages (e.g. Erlam; VanPatten and Cadierno) suggests that drawing learners' attention to pronouns through an explicit focus on linguistic forms contained in the input may be beneficial, especially in the initial stages of learning. This view is consistent with recent cognitive perspectives in linguistics (e.g. Tyler; see also Evans), wherein language is a means of organising, processing and conveying information, and language learning akin to many other cognitive processes.

Despite this interest in the cognitive aspects related to the acquisition of linguistic forms, several scholars (e.g. Cohen 201 la; Hahne, Mueller, and Clahsen; Oxford, Lee, and Park) have highlighted a general lack of research into strategies for learning and using elements of L2 grammar in any language. Cohen (2011a: 689) reports: Although it was the intention of Oxford and Lee (2007) to review the literature on grammar strategy studies, they found that there was such a paucity of studies that instead they wrote a position paper instead [sic] on how grammar strategies had largely been ignored in the research literature.

Our exploratory study aims to contribute to filling this gap, by piloting a mixed-method approach to research into Language Learning Strategies employed by Anglophone university students when engaged in tasks focusing on Italian OPs.

Background to the Study

Language Learning Strategies (LLS) have been researched extensively since the first studies aimed at understanding what makes "Good language learners" (Rubin; Stern). Our preferred definition of LLS is "Thoughts and actions, consciously chosen and operationalized by learners, to assist them in carrying out a multiplicity of tasks from the very onset of learning to the most advanced levels of target-language performance" (Cohen 2011b: 7). In this paper, we focus on LLS that are activated during task performance, i.e. what some scholars refer to as Language Use Strategies, as distinct from Language Learning strategies (Cohen 2011a: 682).

Various studies (e.g., Cohen 2011b; Oxford and Nyikos; Rubin and Thompson) have shown that individual learner variables, such as motivation, self-efficacy, and gender, can influence strategy selection and frequency of use. Individual variation has also been observed in the ways strategies are employed by different learners. For example, more effective language learners appear to use a greater variety of strategies than the less effective, and transfer strategies from task to task (e.g. O'Malley et al. 1985). They also appear better able to plan their learning approach and monitor their success by employing appropriate metacognitive LLS. Vann and Abraham, for example, highlighted the significance of metacognitive strategies in facilitating overall task achievement, and noted that less effective language learners may use LLS inappropriately, being unable to "assess the task and bring to bear the necessary strategies for its completion" (191).

In one of the very few investigations concerned specifically with LLS use in form-focused activities, Chamot, Kupper, and Impink-Hernandez (1988a) conducted a major longitudinal study of Spanish-L2 and Russian-L2 students in the USA. They observed that task difficulty (relative to a student's interlanguage) was important, as less proficient students often used strategies more frequently, although ineffectually, when tackling tasks beyond their level of competence. On the other hand, those who found a task easy could produce correct answers without any deliberate strategy use being evident. "Effective" students were more likely to use metacognitive strategies such as Self-evaluation and Double-check self-monitoring, and to show greater persistence and flexibility in their approach to problem-solving. Task type also mattered, with creative writing tasks eliciting more metacognitive strategies (e.g. Planning), and grammar and vocabulary tasks (e.g. cloze) requiring more frequent use of cognitive strategies, including Deduction, Inferencing and Translation.

While these results are certainly useful, more research is needed in this area to provide an understanding of appropriate strategy selection and use in relation to specific structures of a target language, as well as to refine research methods and tools. For example, researchers (e.g. Chamot et al. 1988a; Cohen 2011a) have found it challenging to interpret observed behaviour univocally, especially when, as often happens, several strategies appear to be operating at once.

Through our study, we sought to inform future, larger-scale research by identifying variables and relationships that warrant further investigation, as well as providing useful analytical tools. More specifically, we aimed to contribute to scholarship on LLS use in form-focused activities by trialling and refining methods for data collection and analysis of cognitive and metacognitive strategies used by learners in tasks focusing on Italian OPs.

This pilot study therefore addressed the following question: In form-focused tasks involving use of Italian OPs, what potential relationships can be identified between students' task outcomes and their use of cognitive and metacognitive strategies in a quantitative (frequency of use) and a qualitative (purpose and type of use) sense?

Method

Our study was conducted in a second-year, second semester course offered as part of the Italian major at an Australian university. The course develops reading and writing skills in particular, and OPs are part of its regular curriculum, in that their forms and rules are explicitly presented, exercises are completed and discussed, and examples of usage from the set readings in the course (short stories and newspaper articles) are examined, in class.

All students in the course were invited to participate and five volunteered to do so (1). However, data collected from one of these were subsequently excluded from consideration, as he appeared unable to articulate his reasoning comprehensively and unambiguously due to insufficient proficiency in English, his L2. The remaining participants were two women (Alpha and Epsilon) and two men (Gamma and Delta), aged between 18 and 28, all Australian-born and native speakers of English.

Our analysis focused on the participants' use of cognitive and metacognitive strategies while performing two complementary tasks. The first required the students to produce OPs in a fill-in-the-gaps activity using a short, previously unseen text entitled "Il caffe di Piero", taken from Nocchi and Naddeo's Grammatica pratica della lingua italiana (see Appendix A). The second involved locating the OPs in four brief passages from the short story "Papa va in TV" by Stefano Benni, and identifying their antecedents (see Appendix B). Although Task 2 did not entail production of OPs, we considered it no less challenging for students than Task 1, and likely to stimulate frequent cognitive and metacognitive strategy use. Given that Task 2 was part of the homework set for all students in the course in that period, the participants were familiar with the content of Benni's story, which had already been discussed in class. Students' familiarity with text content was important given that, while performing the tasks, they only had access to their tables of OP forms, but no dictionaries or other grammar reference resources. This restriction was put in place because we were particularly interested in students' use of their own resources, rather than external ones.

Strategy use data were collected through think-aloud verbal reports. Verbal reports have been widely used in SLA research to study cognitive variables, and are regarded as an effective method of accessing LLS use (e.g. Cohen 2011b; McDonough; see also Cohen 1996 for a review of studies on LLS using verbal reports, and a thorough discussion of advantages and drawbacks of this method).

We informed the participants that the process involved verbal reporting during completion of two tasks on Italian pronouns in the presence of a researcher, and that this would be recorded. The participants were also aware that the object of study was the use of LLS. Following guidelines in the relevant literature (Seliger and Shohamy; Ericsson and Simon), we explained that if they paused during the tasks the observing researcher would invite them to "keep talking" about what they were thinking. As a warm-up activity in thinking aloud we invited the participants to mentally 'walk though' their own house counting the number of windows.

Task verbal reports were recorded outside regular class time by one of the authors, who was unknown to the students prior to the project. We hoped this reassured students that their participation was completely unrelated to course assessment. Recorded reports were transcribed and coded by two of the authors independently, using the LLS taxonomy provided by Chamot et al. (1988a), and reproduced in Table 1.

Initial inter-rater reliability tests, performed on approximately half of the data, pointed to some discrepancies. A Cohen's Kappa test returned a coefficient of 0.66, which could be deemed acceptable; however, we felt that reliability could be improved by increasing clarity in the definition and interpretation of strategy categories. As previously noted, refining analysis categories was one of the aims of this study. Therefore, in the following paragraphs we discuss some of the difficulties we encountered in applying Chamot et al.'s (1988a) taxonomy, and the measures we took to increase its validity and reliability for our specific purposes.

Distinguishing between Self-Evaluative Elaboration and Problem Identification

One difficulty we encountered was in determining whether a student's self-evaluative comment--especially a negative one--manifested: (i) the metacognitive strategy Problem identification, (ii) the cognitive strategy Self-evaluative elaboration aimed at contributing to the problem-solving process, or (iii) no strategy but simply a way of conveying information to the researcher or dealing with embarrassment in her presence. We determined that, if a student said something like "I don't know that word" and then clearly attempted to overcome that specific difficulty by using other strategies, we would code that comment as Problem identification, reserving Self- evaluative elaboration for more casual comments about performance or competence.

Distinguishing between Contextualisation, Elaboration between Parts, and Inferencing

For the participants, an essential part of completing the tasks was identifying and understanding contextual cues, and some discrepancies in our classification arose from uncertainty as to the strategy we were observing when they read the text for meaning. While we used Contextualisation for this, initially, we noticed it was rarefy selected by Chamot et al. (1988a), who seemed to prefer Inferencing. We felt that, while reading the text may be a pre-requisite for Inferencing, it is not the same type of activity, and a separate strategy was involved when students were simply reading ahead seeking semantic cues. We named this extra cognitive strategy Information gathering.

Defining Deduction

Some discrepancies we attributed to what we perceived as lack of clarity in the definition of Deduction. After some discussion, we decided to apply the category of Deduction rather broadly, to all cases in which students seemed to rely on grammatical knowledge in order to understand or produce text. In practice, this meant selecting Deduction every time a student used grammatical metalanguage, unless it was in announcing an answer (such as in stating "That one is a pronoun" during Task 2).

Following this revision of the taxonomy, the same two authors then re-coded the data independently, and the very few discrepancies that arose this time were reconciled so that complete agreement was reached. We then compiled the following quantitative data for each participant: frequency count for each strategy in each task; relative frequency of each strategy in each task (as % of all strategies used by the participant); and task scores, which we used as the measure of students' effectiveness in the tasks. For Task 1, each of the 10 gaps filled correctly scored 1 mark. For Task 2, each of the 12 pronouns located scored half a mark, and correct identification of its antecedent another half mark. We also compared the time each participant took to complete the tasks (in minutes and seconds), in order to ascertain whether more effective students persisted longer in their attempts to complete the tasks successfully. Finally, we examined the transcripts qualitatively, extracting and comparing examples of the ways in which each participant used each strategy.

Data Analysis

Table 2 shows each participant's frequencies of strategy use (as a count and as % of all strategies used by the participant), time on task, and score in Task 1. In completing this fill-in-the-gaps exercise, the more effective the students, the less time they took, and the fewer strategies they used. This inverse relationship is also illustrated in Figures 1 and 2. Gamma, the most effective, completed Task 1 within 4 minutes, while Delta, who found the task difficult, spent over 11 minutes on it, using over three times as many strategies.

Table 3 provides the data on Task 2, which involved locating OPs in a text and identifying their antecedents. Here, the relationships between task scores, time on task, and strategy use are less straightforward than for Task 1. Gamma and Alpha obtained the same score on this task

(11/12), but Alpha spent 33% more time on it than Gamma, and used 42% more strategies. Epsilon spent approximately the same time as Alpha on the task, but used 37% more strategies, and her score was noticeably lower (7/12). Delta spent 43% more time again than Epsilon and used 38% more strategies, while achieving only a slightly higher score (7.5/12). Despite these inconsistencies, as Figures 3 and 4 show, there still appears to be an inverse relationship between strategy frequency and scores: the two less effective students used many more strategies than the other two, and obtained much lower scores.

[FIGURE 1 OMITTED]

[FIGURE 2 OMITTED]

Some of the strategies listed in our taxonomy (Table 1) were completely absent from the transcripts; these included both Metacognitive strategies (Directed attention and Selective attention), and Cognitive strategies (Transfer, Substitution, Note Taking, Grouping, Repetition, Rehearsal and Resourcing (4)).

[FIGURE 3 OMITTED]

[FIGURE 4 OMITTED]

On the other hand, some strategies were employed by all participants rather consistently. Deduction, for example, recorded frequencies ranging between 12.5% and 15% of each participant's total strategy use on Task 1, and between 8.5% and 15% on Task 2. This is not surprising, given that these were form-focused tasks which required applying knowledge of the target language grammar. These results are also consistent with those obtained by Chamot et al. (1988a), who found that Deduction was frequently recorded in cloze-type activities.

Some differences between participants' choices were observed in relation to Information gathering and Elaboration between parts. In Task 1, the two more effective students, Gamma and Alpha, made proportionally greater use of these two strategies than did the less effective students: Gamma recorded 21.9% for both strategies; Alpha: 13% and 14.8%, for Information gathering and Elaboration between parts, respectively; Epsilon: 8.3% and 11.5%; Delta: 7.1% and 8.0%. These were strategies that, together with Deduction, responded most directly to the task demands of understanding the context and identifying pronoun antecedents. Figure 5 shows that higher frequencies for

Information gathering and Elaboration between parts appear to be somewhat related to higher achievement in Task 1.

[FIGURE 5 OMITTED]

[FIGURE 6 OMITTED]

[FIGURE 7 OMITTED]

[FIGURE 8 OMITTED]

Information gathering and Elaboration between parts were also among the most frequently observed strategies in Task 2, again not surprisingly given task demands. However, the relationship between frequency of these two strategies and task scores is less clear than for Task 1, as shown in Figure 6.

Inferencing appears to be used less frequently by the more effective students, as illustrated in Figures 7 and 8. Again this is not surprising, given that Inferencing involves guessing to compensate for incomplete information or understanding--whether informed guessing or otherwise. Delta, who received the lowest score on Task 1, recorded the highest frequency on Inferencing, with 17 instances or 15% of his total strategy count, while Gamma, the most effective participant, only used this strategy 3 times (9.4% of his total).

Qualitative analyses reveal that the participants used Inferencing in different ways, as illustrated by Gamma's and Delta's tackling of the concluding sentences of Il caffe di Piero in Task 1, which were challenging for both students. This is the relevant passage, with the correct answers: Volete mettere con il piacere di preparare la moka, sentirla mentre fa il caffe e sentire il profumo del caffe che riempie la stanza ... poi versarlo nella tazzina, senza zucchero naturalmente. E, finalmente, berlo mentre ti svegli lentamente ... questo si che e un piacere della vita.

Gamma's only incorrect answers in this task were in the last two gaps, where he wrote versarla and berla, as shown in Example 1, below (see Appendix C for transcription symbols): Example 1 [reading] eh sentire il profumo del caffe che riempe la stanza (0.65) poi versar ah ah pt (0.84) eh I think we're still talking about la moka? So again LA? (1.42) poi versarLA? nella tazzina (0.68) eh senza zucchero naturalmente. E finalmente ber LA? We're still talking about la moka? berLA mentre ti svegli lentamente

Gamma seems to have interpreted la moka as a variety of coffee, like Arabica, which also occurred in the text. His Inferencing here appears to be informed guessing at the meaning of la moka--based on the context, and drawing on a good overall understanding of the text. He appears very confident, as there is nothing that suggests he considered other options. However, he may have been aware that this was a gap in his knowledge, and have decided not to waste time and effort on it but guess and move on.

Delta also arrived at wrong answers for these two gaps--versarle and berle--but through a process that involved some random guessing, as he confessed, on both the meaning of versare and whether it took a direct or indirect object. For example, he changed his initial answer of versarla to versarle as a result of thinking "so I'm going with, oh, is it direct? I reckon it's indirect 'cos I'm taking a complete stab at this one, so let's go versarLE".

Overall, while Delta tended to use Inferencing throughout the tasks mostly in the form of random guessing to cover for lack of knowledge (especially of word meanings), the more effective participants--Gamma and Alpha--tended to use it for informed guessing, drawing on their understanding of other parts of the text and links within the text, and often making explicit their justification for the guess. For example, they often applied Inferencing when investigating relationships between words, and when trying to interpret sentences or the whole text, rather than in dealing with meaning of single words. Even much of Epsilon's Inferencing was expressed in ways such as: "I think she's [Lucilla's] speaking", "I think they're talking about il caffe", "I'm assuming that's from, that's the first part of bere? That I need to add the pronoun to the end of?", although she also made some "stabs" like Delta.

The strategies of Deduction and Translation were often employed together, and also in combination with Elaboration between parts. Like Inferencing, these strategies were used in different ways by our participants, due to their differing levels of understanding of the texts and degrees of attention to the importance of linking form and meaning. Alpha and Gamma, the more effective students, tended to spell out a precise and detailed translation of a fragment, on the basis of which they then applied a grammatical rule. In Example 2, below, Gamma was evidently concerned to provide himself with a translation that satisfied him as making complete sense of the use of the indirect third person singular feminine OP: Example 2 is ehm the LE is obviously the indirect object pronoun of (.) the-third person feminine? Eh Zitta stanno inquadrato- inquadrando papa dice Lea senza prestarle attenzione. (1.087) So she is NOT paying attention to- Lea is not paying attention to Lucilla. (1.754) (I think).

Alpha, too, always seemed to want to have a precise meaning to work with. Her approach to the same pronoun was similar to Gamma's: Example 3 (Senza prestarle) attenzione OK (2.2) So urn le is (2.9) [Researcher: What are you thinking there?] Um trying to (*) that's- it's obviously a p- (*) a pronoun of some sort, I'm just trying to figure out what it refers to (2.4) um (2.9) oh I see ok Ok So she's just (2.0) the girl's picked up the phone (*) and it's a journalist (*) blablabla but Lea (*) is too busy watching her husband getting killed on tv so she's not paying attention to the daughter. SO I think that is (0.5) urn (1.5) that's referring to the daughter, to Lucilla. She's not paying to HER any attention so it's (indirect) (2.7) OK\. That's all (0.9) finished.

Another case in which both Gamma and Alpha were attentive to getting at the exact meaning was in relation to the line "Si, e un bell'uomo ma non so se gli somiglia davvero" in another passage included in Task 2. Both grappled with whether the OP gli had to do with Augusto (the Papa character) resembling the actor Depardieu, or vice versa.

These observations suggest that both Gamma and Alpha expect that understanding meaning precisely and completely--and at sentence or even whole text level--will be essential to getting correct answers. And quite often they did not spell out the application of the rule (i.e. they did not make explicit a Deduction strategy), as their translation was enough to render clearly why they chose the answer they did.

However, Alpha encountered more obstacles to understanding than Gamma did. When unable to achieve the degree of understanding that made her comfortable, she sometimes lost confidence, and either gave up or resorted to random guessing, usually after expressing disappointment or frustration (coded as Self-evaluative elaboration), as shown in Example 4: Example 4 (e l'unica che) [reading] (3.59) [Researcher: Just keep talking, whatever comes into your mind there that's fine] I'm not quite sure what this next sentence is (*) about? um (1.3) XX (4.4) (tiene sveglio) e (3.57) [Researcher: You can read it aloud if you like, if it'll help] (1.30) E l'unica che something tiene sveglio (1.6) I don't know what tiene (0.5) I get mental blanks about these things all the time! Um. (5.9) No. I don't know what that sentence means. [Researcher: Uhuh Continue?] [Later, when she comes back to it] OK so (*) This one up here I just don't know what this sentence means and I don't know what it's referring to. So (2.28) E l'unica che (*) s- something tiene (*) sveglio. (7.71) Ok\ (*) I'm going to go out on a limb and guess that it's still referring to questa miskela from the previous sentence. I'm just gonna put la in there everywhere (1.5) Because I don't know what these verbs mean. I don't know what it's about. I don't know whether it's direct or indirect. [A bit agitated] So I'm just going to (*) guess (0.8). That it's LA for both of those. And I'm done. With that one. [Researcher, in a cheerful voice: Ok!] It's really hard to do when you don't actually know what (*) the sentences mean. Heh? [Researcher: Um, Fully understand] Yep. [Researcher: Alright, well, leave that one for now and go on to (*) the next one]

Indeed, a feature of Alpha's verbal report is her engagement with the text and task on a deeper level than Gamma, manifested through her frequent reactions to and reflections on textual meaning (classified under the strategy of Elaboration) as well as on her own performance in the task (classified under Self-evaluative elaboration). This partly accounts for her spending as much time on Task 2 as Epsilon--that is, 33% more time than Gamma--while achieving the same high score as Gamma. Example 5 illustrates Alpha's engagement with textual meaning: Example 5 I'm just reading ahead. Reading the next sentence. I'm trying to figure out what deteinato means. I don't know, Is that the equivalent of decaf but for tea? (3.00) Why would you do that to tea? That's a horrible thing to do to tea; it ruins it!

Compared with Gamma and Alpha, Epsilon and Delta seemed to be handicapped by not only greater difficulty in understanding the texts but also less concern for precision in understanding them and applying rules. They used Translation frequently, but usually on fragments too short to allow them to establish a sufficient understanding of relationships between words. And, although Epsilon exhibited a very good declarative knowledge of the pronoun forms and rules, in her frequent explicit references to them (Deduction strategy) she often demonstrated a lack of concern that understanding textual meaning is important for applying them correctly.

For instance, in Example 6, Epsilon seemed to apply her rules somewhat mechanically, by focusing on morphology--the gender and number of the OP--rather than combining this information with textual meaning. She appeared to see the problem as one of identifying anything singular and masculine nearby as the antecedent of the pronoun lo in la moglie lo rimprovera. She first considered un sacco di lettere (because un sacco is singular masculine), then reconsidered it, sensibly, and opted for Mario. Overall in this, as in other cases, her interpretation of 'what a pronoun refers to' appeared to encompass any noun that might be loosely associated with it in some way.

As for metacognitive strategies, we found little use of them overall, but substantially more use by Epsilon and Delta than by the more effective students Gamma and Alpha. This undoubtedly contributed to Epsilon's and Delta's greater time on both Task 1 and Task 2. The metacognitive strategies used by our participants were primarily Self- monitoring and Self-management.

Notably, only Epsilon adopted an overall Planning strategy to establish her approach to a task at the start, which we would have recommended strongly if facilitating the activity in class. She read most of the way through the text of Task 1 to get the gist, before going back to the start and beginning to work on the pronouns. Similarly, in Task 2, Epsilon read each excerpt that made up the task first--with some translation as she went--before re-reading it in order to work on the pronouns. There was no evidence that the more effective students Gamma and Alpha did this: they seemed usually to read forward only as far as each gap (in Task 1) or each pronoun (in Task 2), and deal with it before reading further. Example 6 uhm [plenty of letters] fan letters love letters Mario is talking there uhm oh I think the wife la moglie is referring to the che ricevera un sacco so this (*) sac of letters bag of letters 'cos LO is uhm singular masculine uhm (0.22) unless she's referring to Mario uhm actually no yeah I think she's referring to Mario there yeah. So I'm gonna change that one to referring to Mario; that makes more sense

Epsilon was also the only one to exhibit extensive Double-check self- monitoring (sometimes in combination with Self-management), which she used systematically and productively by deliberately leaving some answers open and returning to them later, and even going back to Task 1 after completing Task 2, to review her work completely and revise her answers where she could. Her transcript was rich with statements like "actually (*) I'm just realising one down the bottom is wrong uhm I'll just (*) come back to it" and "[I think I'll go back to] the one I skipped at the top" and "I'm just unsure on a couple of these, I'm gonna go back". An example of Epsilon's effective use of a Self-monitoring strategy is in her dealing with la moglie lo rimprovera in Example 6, above. Coming back to it in her second pass of the task, and then taking the time to mull over the problem, meant that she let a second option come to mind as the correct answer.

While the most effective student, Gamma, gave very little evidence of metacognitive strategies in his verbal report, there were pauses during which he may have used them without verbalising the processing. For instance, following his dealing with the OP le in the sentence dice Lea senza prestarle attenzione, in Example 2, Gamma was silent for 1.75 seconds, after which he said "I think" under his breath. During that pause, it is possible that Gamma employed Double-check self-monitoring, or indeed other strategies, but automatically rather than consciously, and therefore without seeing them as part of his problem-solving process.

Discussion and Conclusion

With respect to our research question, our data suggest an inverse relationship between task results, on one hand, and strategy frequencies and time on task, on the other--in contrast with Chamot et al.'s (1988a) findings. Given the small number of participants, this apparent relationship cannot be generalised. Nevertheless, these results are interesting as they suggest avenues that could be explored further with a larger pool of participants.

Our data also contrast with those of Chamot et al. (1988a) in that our more effective students did not exhibit a broader repertoire of strategies, or more frequent or appropriate use of metacognitive strategies, than the less effective ones. However, they did make proportionally greater use of the cognitive strategies that we consider most relevant to the form-focused tasks used in this study: Information gathering (reading for meaning), Elaboration between parts (finding links between different elements contained in the text), and Deduction. The usefulness of Elaboration between parts as a way of gathering contextual cues and therefore facilitating comprehension in cloze tasks was also recognised in Chamot et al.'s (1988b: 37) study, as was Information gathering, which they classified as a type of Inferencing.

Interestingly, our more effective students tended to use Deduction, Translation and Inferencing in different ways from those of the less effective, by basing their informed guessing on a deeper understanding of textual meaning, and of the relationship between linguistic forms and meaning. Conversely, the less effective students, Epsilon and Delta, tended to employ the same strategies somewhat mechanically, and with limited attention to semantic elements, which often undermined their success in applying the strategy.

As for the metacognitive strategies, it was interesting to note that it was Epsilon who made most use of them, and usually in highly appropriate ways. This is rather surprising, given Epsilon's task scores. Indeed there are several cases where, if Epsilon had sought to articulate in detail what a sentence meant, she might well have obtained correct answers despite the constraints of her proficiency level. She did seem to struggle, however, with the cognitive load of paying attention to both form and meaning.

The most effective student, Gamma, made least use of metacognitive strategies. This result contrasts with Chamot et al.'s (1988a) observation that frequent use of metacognitive strategies, especially Double-check self-monitoring, was a trait of high-achieving students. Gamma's apparent lack of use of other metacognitive strategies such as Self-evaluative elaboration and Self-monitoring may be related to the ease with which he tackled these tasks. Delta, who encountered the most difficulty, used both strategies much more frequently.

As Chamot et al. (1988a) also observed, the level of challenge posed by a task is an important factor. Students who find a task within their level of expertise may not need to employ strategies at all. It is also possible that they use the same cognitive and metacognitive processes as the weaker students, but without being aware of them, having internalised them through practice (see Anderson for a thorough discussion of the differences between Controlled and Automatic processing). In that case, as O'Malley and Chamot noted, such processes can no longer be considered strategies.

An implication for future research is that highly proficient students may exhibit relatively low strategy use, unless the tasks pose a degree of challenge commensurate with their knowledge and skills. Investigating cognitive and metacognitive strategies may be most productive at lower-intermediate levels of linguistic competence, where tasks present students with several 'problems' to solve, and some appropriate strategies have been developed but have not yet become automatic. This will ensure that students are required to activate strategies frequently and consciously, which makes them available for observation. On the other hand, as our results suggest, when a task is beyond a student's level of competence, high strategy frequencies and variety do not necessarily lead to positive task outcomes. Providing access to external resources, such as dictionaries, might allow weaker students to unlock some of their cognitive resources for further processing, leading to patterns of strategy use that differ from what we observed. As previously noted, our participants did not have access to external resources. As a result, relative strategy frequencies (both for Resourcing and other strategies), time on task, and task score, may have been significantly affected.

Finally, it is likely that individual personality traits may influence a student's propensity for selecting and using certain strategies, as our data suggest in relation to self-evaluative comments. Gender may be also a factor, since research (e.g. Tannen) suggests that women tend to share personal views and aim for emotional connection in spoken interaction more than men. In our study, the two female participants, Alpha and Epsilon, indeed made personal comments frequently.

In summary, our results point to areas for further investigation that could yield interesting results. These include:

* Relationships between strategy frequency, time on task and task scores, while controlling for task difficulty;

* Qualitative differences in the use of Deduction, Translation and Inferencing by more effective and less effective learners;

* Influence of personality traits and gender differences on the use of metacognitive strategies involving self-evaluation and elaboration (e.g. emotional reactions to texts).

In addition to identifying relevant variables, this project allowed us to test and refine the LLS taxonomy we borrowed from Chamot et al. (1988a). The changes and clarifications we applied increased its validity and reliability for our purposes, and we are confident the revised taxonomy is sufficiently robust for future studies. Nevertheless, some uncertainty inherent in matching participants' words and behaviour to strategy definitions remains, especially when different strategies are employed concurrently. Thorough checks for strong inter-rater reliability must be put in place to reduce such uncertainty.

The degree to which a think-aloud verbal report accurately reflects the student's strategic thinking is also questionable, especially in the presence of a researcher. For example, the think-aloud process itself may render some cognitive processes conscious and verbalised that would otherwise be automatic and therefore not be considered strategies in the sense used in LLS research. In some cases, our participants seemed to address the researcher out of a desire to communicate, by sharing positive feelings about the text, or attempting to project a positive self-image despite embarrassment at their poor performance. Conversely, students may neglect to verbalise their thoughts at times, due to lack of experience with think-aloud protocols, lapses of attention, or complete internalisation of mental processes, as suggested by Gamma's pauses.

In order to address some of the above limitations, we recommend that future studies employ a mixed-mode approach which includes quantitative analysis of larger data sets--to explore the statistical significance of correlations between strategy frequencies and individual learner variables, for example--as well as systematic qualitative analysis of think-aloud data. Such analysis could be further strengthened by adding retrospective reports collected in post-think-aloud interviews.

In conclusion, despite its limited scale, we believe that our pilot study is a positive contribution toward a better understanding of LLS use for learning and practising elements of the target language grammar. Our experience allowed us to identify variables and relationships that warrant further investigation, and to refine our analysis tools, and represents a useful starting point for larger-scale projects on LLS use in tasks focusing on Italian OPs.

Notes

(1) Approval to carry out this study was obtained from the Ethics Committees at the two Universities in which the authors work. Pseudonyms have been used to protect the participants' identities.

(2) While the strategies listed in Table 1 were available to the researchers during the coding phase, not all of these strategies were indeed found in the data.

(3) The frequency of the metacognitive strategy Planning is not shown as % of each participant's total strategies, nor is it counted in those totals, as it is a strategy likely to be used at most once, at the start of a task, and not during task execution. In our case, it was only employed by Epsilon.

(4) In the case of Resourcing, it is worth noting that our participants did not have access to dictionaries or other external sources of assistance, which certainly influenced observed frequencies.

Works Cited

Anderson, John R. Learning and Memory: An Integrated Approach. New York: Wiley, 1995. Print.

Benni, Stefano. Liultima lacrima. Milano: Feltrinelli, 1996. Print.

Berretta, Monica. "Per uno studio dell'apprendimento dell'italiano in contesto naturale: Il caso dei pronomi naturali atoni". Li'apprendimento spontaneo di una lingua seconda. Giacalone Ramat, Anna (a cura di). Bologna: Il Mulino, 1986, 329-52. Print.

Chamot, Anna U., Lisa Kupper, and M. Impink-Hernandez. "A Study of Learning Strategies in Foreign Language Instruction: Findings of the Longitudinal Study." (1988a). Web. 5 July 2015. <http://eric.ed.gov/?id=ED352823>

Chamot, Anna U., Lisa Kupper, and M. Impink-Hernandez. "A Study of Learning Strategies in Foreign Language Instruction. The Third Year and Final Report." (1988b). Web. 5 July 2015. <http://eric.ed.gov/?id=ED352825>

Cohen, Andrew D. "Verbal Reports as a Source of Insights into Second Language Learner Strategies." Applied Language Learning 7.1&2 (1996): 5-24. Print.

Cohen, Andrew D. "Second Language Learner Strategies". Handbook of Research in Second Language Teaching and Learning. Hinkel, Eli (ed.) London: Routledge, 2011a, 681-98. Print.

Cohen, Andrew D. Strategies in learning and using a second language. London: Routledge, 2011b. Print.

Costi, Elena. "L'uso dei pronomi atoni nella scuola primaria: confronto tra parlanti nativi e non nativi". Thesis. Universita degli Studi di Modena e Reggio Emilia, 2007. Web. 5 July 2015. <http://www.comune.re.it/ download / interlingua/ tesi_elisa_costi.pdf>

Ericsson, K. Anders and Herbert A. Simon. Protocol Analysis: Verbal Reports as Data. Cambridge, MA: MIT, 1993. Print.

Erlam, Rosemary. "The Effects of Deductive and Inductive Instruction on the Acquisition of Direct Object Pronouns in French as a Second Language". The Modern Language Journal 87.1 (2003): 242-60. Print.

Evans, Vyvyan. The Language Myth: Why Language Is Not an Instinct. Cambridge: Cambridge UP, 2014. Print.

Hahne, Anja, Jutta L. Mueller, and Harald Clahsen. "Morphological Processing in a Second Language: Behavioral and Event-related Brain Potential Evidence for Storage and Decomposition." Journal of Cognitive Neuroscience 8.1 (2006): 121-34. Web. 5 July 2015. <http://www.ncbi.nlm. nih.gov/pubmed/16417688>

Hales, Dianne R. La Bella Lingua: My Love Affair with Italian, the Worldis Most Enchanting Language. New York: Broadway, 2009. Print.

McDonough, Steven H. Strategy and skill in learning a foreign language. New York: Oxford UP, 1995. Print.

Nocchi, Susanna e Ciro Massimo Naddeo. Grammatica pratica della lingua italiana: esercizi, test, giochi. Firenze: Alma, 2002. Print.

O'Malley, J. Michael, and Anna U. Chamot. Learning Strategies in Second Language Acquisition. Cambridge: Cambridge UP, 1990. Print.

O'Malley, J. Michael, Anna U. Chamot, Gloria Strewner-Manzanares, Lisa Kupper, and Rocco P. Russo. "Learning Strategies Used by Beginning and Intermediate ESL Students." Language Learning 35.1 (1985): 21-46. Web. 5 July 2015. <http://doi: 10.1111/j.1467-1770.1985.tb01013.x>

Oxford, Rebecca L., Kyoung Rang Lee, and Gloria Park. "L2 Grammar Strategies: The Second Cinderella and beyond." Language Learner Strategies: 30 Years of Research and Practice. Cohen, Andrew D. and Ernesto Macaro (eds.) Oxford: Oxford UP, 2007, 117-39. Print.

Oxford, Rebecca and Martha Nyikos. "Variables Affecting Choice of Language Learning Strategies by University Students." The Modern Language Journal 73.3 (1989): 291-300. Web. 5 July 2015. <http://doi. org/10.1111/j.1540-4781.1989.tb06367.x>

Pona, Alan. "I pronomi clitici nell'apprendimento dell'italiano come L2: Il clitico 'si' nelle varieta di apprendimento." Annali Online Lettere 4.2 (2009), 14-39. Web. 5 July 2015. <http://annali.unife.it/lettere/article/ viewFile/190/139>

Rubin, Joan. "What the 'Good Language Learner' Can Teach Us." TESOL Quarterly 9.1 (1975), 41-51. Web. 5 July 2015.<http://doi. org /10.2307 / 3586011>

Rubin, Joan and Irene Thompson. How to Be a More Successful Language Learner. Boston: Heinle & Heinle, 1982. Print.

Seliger, Herbert W. and Elana Shohamy. Second Language Research Methods. Oxford: Oxford UP, 1989. Print.

Stern, Hans H. "What Can We Learn from the Good Language Learner?" Canadian Modern Language Review 31.4 (1975), 304-18. Print.

Tannen, Deborah. You Just Dom't Understand: Women and Men in Conversation. New York: Morrow, 1990. Print.

Tyler, Andrea. Cognitive Linguistics and Second Language Learning: Theoretical Basics and Experimental Evidence. New York: Routledge, 2012. Print.

Vann, Roberta J. and Roberta G. Abraham. "Strategies of Unsuccessful Language Learners." TESOL Quarterly 24.2 (1990), 177-98. Web. 5 July 2015. <http://doi.org/10.2307/3586898>

VanPatten, Bill, and Teresa Cadierno. "Explicit Instruction and Input Processing." Studies in Second Language Acquisition 15.2 (1993), 225-43. Print.

Zorzi, Daniela. "I pronomi clitici nell'uso degli studenti anglofoni". Le Lingue del Mondo LII.3-4 (1987), 41-53. Print.

Appendix A

Task 1, based on the text "Il caffe di Piero"

Pronomi diretti e indiretti

Completa con i pronomi diretti o indiretti Il caffe di Piero

Che bellezza! Il mio primo caffe della giornata! Ho preso la moka, ___ ho riempit ___ di acqua, ho messo il filtro e poi ho aggiunto il mio caffe preferito: Arabica._piace questa miscela, ___ bevo ogni giorno. E l'unica che ___ tiene sveglio e ___ fa affrontare bene la giornata. La mia ragazza invece e una salutista, dice che il caffe non ___ fa bene, che ___ rende nervosa e beve solo te deteinato. Si! Avete capito bene: te deteinato, un'offesa all'italianita! Volete mettere con il piacere di preparare la moka, sentir ___ mentre fa il caffe che riempie la stanza ... poi versar ___ nella tazzina, senza zucchero naturalmente. E finalmente, ver ___ mentre ti svegli lentamente ... questo si che e un piacere della vita!

[ILLUSTRATION OMITTED]

Appendix B

Task 2, based on excerpts from "Papa va in TV" by Stefano Benni (1994)

Individua i pronomi oggetto in questi brani del racconto "Papa va in TV" e, per ciascun pronome, il nome a cui si riferisce. 1. A che ora e il collegamento televisivo? Tra cinque minuti, piu o meno. Allora possiamo accendere. Il telecomando lo tengo io--dice Lucilla. Lucilla non fare la prepotente. Papa me lo fa sempre tenere. 2. Si e pettinato all'indietro, come gli avevo detto. Mi sa che ricevera un sacco di lettere di ammiratrici--dice Mario. La moglie lo rimprovera con lo sguardo. Ecco, si siede. Guarda che bel primo piano. Vecchio Augusto!--dice Mario un po' commosso--chi l'avrebbe mai pensato! 3. L'ho legata troppo stretta?--chiede il tecnico. No, no, va benissimo--risponde Augusto. Se vuole un consiglio, quando arriva la scarica, tenga la testa giu. Cosi non si vedono le smorfie ... Ma io vorrei che a casa mi vedessero bene. 4. Hai sentito--dice Mariella, tutta eccitata--l'ha paragonato a Depardieu! Non sei contenta? Beh, si, e un bell'uomo ma non so se gli somiglia davvero ...--dice Lea, timida. Squilla il telefono. Mamma--dice Lucilla--e un giornalista. Chiede che cosa stiamo probando ... Zitta, stanno inquadrando papa--dice Lea senza prestarle attenzione. Appendix C Transcription symbols CAPITAL loudness (word) whisper (.) micro-pause (.3) pause (measured in seconds) [reading] description, comment, added information [...] omission--part of transcript is not provided \ downward intonation ? upward intonation h... / hhhh audible expiration / laughter XX inaudible number of syllables

Antonella Strambi

Flinders University

Claire Kennedy

Griffith University

Wendy Dekker

Flinders University Table 1: Strategy categories from Chamot et al. (1988a, pp. 11-19 to 11-29) Strategies (2) Description (paraphrased or directly cited from Chamot et al. 1988a, pp. 11-19 to 11-29) Metacognitive Planning Previewing task; identifying useful strategies; generating a plan Directed attention Consciously directing one's attention to task Selective attention "Deciding to attend (or attending) to specific aspects of language input or situational details" Self-management Knowing what works best for oneself, and ensuring that these best con- ditions are established Self-monitoring "Checking, verifying, or correcting one's comprehension or perfor- mance in the course of a language task" Includes Double-check self-monitoring, i.e. returning to an unsolved problem Problem "Explicitly identifying [what] hinders identification successful accomplishment of [the] task" Cognitive Repetition Repeating words to keep them in mind while working something out Rehearsal "Rehearsing language presumed to be needed for a task, with attention to meaning" Resourcing Using reference resources, e.g. dictionary, textbook etc. Grouping "Ordering, classifying, or labelling material" Note taking "Writing down key words and concepts in abbreviated ... form" Deduction "Consciously applying learned or self-developed rules" Substitution "Selecting alternative approaches" Contextualisation "Placing a word or phrase in a meaningful language sequence" Elaboration "Relating new information to prior knowledge; relating different parts of new information to each other; making meaningful personal associations to information presented". Includes: --using personal, academic or world knowledge --elaboration between parts (connecting different parts of the same task) --brainstorming through questions --self-evaluative observations Summarisation Summarising or translating (not verbatim) Translation Rendering information provided in the task into a different language Transfer "Using previously acquired linguistic knowledge to facilitate a language task" Inferencing "Using available information: to guess the meanings or usage of unfamiliar language ...; to predict outcomes, or to fill in missing information" Table 2: Strategy use and performance for each participant, Task 1 FREQUENCY OF STRATEGY USE BY STUDENT STRATEGY GAMMA ALPHA N % N % Cognitive strategies: Elaboration between parts 7 21.9 8 14.8 Information gathering 7 21.9 7 13.0 Deduction 4 12.5 8 14.8 Inferencing 3 9.4 7 13.0 Repetition 4 12.5 3 5.6 Elaboration (e.g. engaging 0 0.0 7 13.0 prior knowledge, questioning) Translation 2 6.3 3 5.6 Self-evaluative elaboration 0 0.0 3 5.6 Summarising 1 3.1 I 1.9 Metacognitive strategies: Self-monitoring 3 9.4 2 3.7 Problem identification 0 0.0 5 9.3 Self-management 1 3.1 0 0.0 Planning (3) 0 0 TOTAL STRATEGY 32 100 54 100 FREQUENCY TASK SCORE (out of 10) 8 7 TIME ON TASK (min:sec) 3:55 6:35 FREQUENCY OF STRATEGY USE BY STUDENT STRATEGY EPSILON DELTA N % N % Cognitive strategies: Elaboration between parts 11 11.5 9 8.0 Information gathering 8 8.3 8 7.1 Deduction 13 13.5 17 15.0 Inferencing 13 13.5 17 15.0 Repetition 7 7.3 13 11.5 Elaboration (e.g. engaging 3 3.1 13 11.5 prior knowledge, questioning) Translation 11 11.5 3 2.7 Self-evaluative elaboration 6 6.3 10 8.8 Summarising 1 1.0 3 2.7 Metacognitive strategies: Self-monitoring 18 18.8 16 14,2 Problem identification 3 3.1 3 2.7 Self-management 2 2.1 1 0.9 Planning (3) 1 0 TOTAL STRATEGY 96 100 113 100 FREQUENCY TASK SCORE (out of 10) 4.5 2.5 TIME ON TASK (min:sec) 10:53 11:18 Table 3: Strategy use and performance for each participant, Task (2) FREQUENCY OF STRATEGY USE BY STUDENT STRATEGY GAMMA ALPHA N % N % Cognitive strategies: Elaboration between parts 6 11.3 7 9.3 Information gathering 10 18.9 9 12.0 Deduction 8 15.1 9 12.0 Inferencing 1 1.9 5 6.7 Repetition 8 15.1 6 8.0 Elaboration (e.g. engaging prior 2 3.8 10 13.3 knowledge, questioning) Translation 6 11.3 5 6,7 Self-evaluative elaboration 1 1.9 5 6.7 Summarising 3 5.7 6 8.0 Metacognitive strategies: Self-monitoring 6 11.3 5 6.7 Problem identification 2 3.8 6 8.0 Self-management 0 0.0 2 2.7 Planning (3) 0 0 TOTAL STRATEGY 53 100 75 100 FREQUENCY TASK SCORE (out of lO) 11 11 TIME ON TASK (min:sec) 9:01 12:00 FREQUENCY OF STRATEGY USE BY STUDENT STRATEGY EPSILON DELTA N % N % Cognitive strategies: Elaboration between parts 14 13.6 21 14.8 Information gathering 12 11.7 9 6.3 Deduction 12 11.7 12 8.5 Inferencing 10 9.7 14 9.9 Repetition 8 7.8 16 11.3 Elaboration (e.g. engaging prior 9 8.7 13 9.2 knowledge, questioning) Translation 6 5.8 14 9.9 Self-evaluative elaboration 6 5.8 12 8.5 Summarising 3 2.9 5 3.5 Metacognitive strategies: Self-monitoring 12 11.7 14 9.9 Problem identification 5 4.9 4 2.8 Self-management 6 5.8 8 5.6 Planning (3) 1 0 TOTAL STRATEGY 103 100 142 100 FREQUENCY TASK SCORE (out of lO) 7 7.5 TIME ON TASK (min:sec) 12:12 17:32