The effect of focused communication tasks on instructed acquisition of English past counterfactual conditionals

One important controversy connected with the effectiveness of grammar teaching seems to have been resolved as there is ample empirical evidence testifying to the positive effect of form-focused instruction on second language acquisition (Nassaji & Fotos, 2004; Norris & Ortega, 2000; Spada, 1997, 2010). Nevertheless, there are still a number of problems open to debate and awaiting concrete solutions, such as how to establish connections between form and meaning and find the best way to teach grammar for implicit knowledge, which, in the opinion of most SLA researchers (Ellis, 2006a, p. 95) and according to numerous theoretical positions, is a key driver of linguistic competence. One of the options available to language educators is to employ focused communication tasks, which “are designed to elicit production of a specific target feature in the context of performing a communicative task” (Ellis, 2001, p. 21). The aim of the study reported in this article was to explore the effect of focused communication tasks on the instructed acquisition of English past counterfactual conditionals when compared with contextualized practice activities. The results of two types of intervention were measured employing a number of data collection instruments with a view to tapping both the explicit and implicit knowledge of the participants of the study. Both types of instructional treatment were equally effective in helping learners develop the explicit knowledge of past unreal conditionals, but when it comes to the implicit knowledge of the aforementioned structure, the group instructed by means of focused communication tasks outperformed the other experimental group and the control group, as evi-

The central role of grammar in the language curriculum was relatively unquestionable for many years and it would have been unthinkable to imagine language instruction without grammar until the late twentieth century when new theories and approaches started to emerge.They were concerned with the roles of explicit and implicit learning and tried to establish whether learning occurs through conscious manipulation or unconscious processes taking place during exposure to input (e.g., N. Ellis, 1994).With the advent of Krashen's (e.g., 1981) input hypothesis, form-focused instruction was no longer considered necessary for language acquisition.Classroom procedures were to resemble naturalistic contexts on the basis of the assumption that language can be acquired from exposure to comprehensible input only.The communicative approach advocated the exclusive use of meaning-focused activities in a foreign language classroom (e.g., Prabhu, 1987).With time, however, it turned out that the rejection of formal instruction was premature.The complete abandonment of grammar teaching proved unsuccessful as learners were unable to achieve high levels of grammatical accuracy even though they had plentiful opportunities for meaningful practice (e.g., DeKeyser, 1998;Doughty, 1991;Lightbown, 1998;Lightbown & Spada, 1990;Norris & Ortega, 2000;Robinson, 1996;Spada & Lightbown, 1993;Swain, 1985).The role of grammar has had to be reconsidered and there are at least four important reasons which support instruction in the formal aspects of the language (Nassaji & Fotos, 2004).First of all, the idea that noticing and attention are necessary for second language acquisition was argued by the proponents of the noticing hypothesis (Schmidt, 1990(Schmidt, , 2001) ) and has been supported by numerous researchers (DeKeyser, 1998;Doughty, 2001;Ellis, 2001).It has been proved that noticing is necessary for learners to attend to form in the input; otherwise input is processed for meaning only and the structure may be ignored.Secondly, the research has provided evidence for the existence of developmental sequences which learners have to undergo, which resulted in the teachability hypothesis (Pienemann, 1984) suggesting that while certain developmental sequences are fixed and cannot be altered by grammar teaching, other structures can benefit from instruction any time they are taught.Instruction is therefore believed to be able to influence sequences of development if learners are ready to move to the next developmental stage of linguistic proficiency (Lightbown, 2000).Thirdly, there is a large body of research pointing to the inadequacies of teaching approaches where the focus is primarily on meaning-focused communication, and grammar is not addressed.Early immersion programmes in Canada proved unsuccessful in terms of grammar accuracy, and instruction in the formal aspects of the language was necessary to develop high levels of accuracy in the target language.Finally, there is plenty of evidence for positive effects of grammar instruction from classroom research.The meta-analysis by Norris and Ortega (2000) has found that explicit instruction (presenting the structure, describing and exemplifying it, and giving rules for its use) results in substantial and long-term gains in the learning of target structures in comparison to implicit instruction which usually entails communicative exposure to the target form only. New perspectives on grammar instruction in foreign language classrooms, based on emerging theories of language learning and teaching, introduced new dimensions of form-focused instruction and generated a number of novel teaching options.As Burgess and Etherington (2002, p. 433) stated: "grammar is being rehabilitated . . .and recognised for what it has always been . ..: an essential, inescapable component of language use and language learning."

Focused Communication Tasks
Among a plethora of controversial issues connected with form-focused instruction there is the question about instructional options which can foster the process of learning grammatical structures so that students develop high levels of accuracy in the target language.There is broad consensus that learners need to have opportunities to encounter, process, and use the new forms in their form-meaning relationships so that they can become part of their interlanguage (Ellis, 2002;Larsen-Freeman, 2003;Nassaji, 2000;Nassaji & Fotos, 2010;Pawlak, 2005;Spada, 2010).One of the principles that Ellis proposes states that "learners need the opportunity to participate in communicative activities to develop implicit knowledge.Thus, communicative tasks need to play a central role in instruction directed at implicit knowledge " (2005b, p. 13).It cannot be denied that recent years have seen a growing interest in the use of tasks in the process of foreign language teaching (see e.g., Ellis, 2003;Nunan, 1991Nunan, , 2004;;Skehan, 1996Skehan, , 2003)).Various tasks are believed to contribute to the acquisition of language forms and promote successful language production.Nevertheless, as Nassaji (2000) observes, many second language acquisition researchers argue that activities focusing solely on message conveyance can be inadequate to foster the development of accurate language use and that some focus on form is necessary in communicative classroom contexts (e.g., DeKeyser, 1998;Doughty, 1991;Lightbown, 1998;Lightbown & Spada, 1990;Norris & Ortega, 2001;Pawlak, 2006;Robinson, 1996;Spada & Lightbown, 1993;Swain, 1985).
In order to be able to explore the effectiveness of focused communication tasks in instructed second language acquisition of grammatical structures, a precise definition of a task is indispensable.A number of different proposals have been suggested (e.g., Ellis 2003;Long, 1983;Prabhu, 1987;Skehan, 1998) and the most general distinction was offered by Nunan (2004), who distinguished between real world target tasks and pedagogical tasks, where the main difference is whether the task is used beyond or in the classroom.As far as a pedagogical task is concerned, various definitions have been constructed as well (e.g., Bygate, Skehan, & Swain, 2001;Ellis, 2003;Skehan, 1998).Nunan's (2004, p. 4) definition, which describes a pedagogical task as "a piece of classroom work which involves learners in comprehending, manipulating, producing or interacting in the target language while their attention is principally focused on mobilizing their grammatical knowledge in order to express meaning, and in which the intention is to convey meaning rather than to manipulate form," is in line with Ellis's (2003) definition in the sense that both of them acknowledge the interrelatedness of form and meaning and their importance for the effective expression of various communicative functions.There are two main arguments for the implementation of communication tasks in classroom teaching.According to Griggs (2005, p. 407), they provide a favourable learning situation, because "they place learners in the centre of the learning process by creating an interactional framework in which they solve language problems in order to fulfill communicative needs."Having been exposed to the task which contains a gap between different sources of information, students usually work in groups or in pairs to partake in communicative interaction.Second of all, communication tasks help bridge the gap between language learning in an educational context and actual language use in the real world (Nunan, 1991), and therefore they are believed to contribute incidentally to learners' linguistic development, with regard to both fluency and accuracy (Ellis, 1997).
Communication tasks can be divided into two types: unfocused and focused (Ellis, 1997(Ellis, , 2003;;Nunan, 2004).In the case of unfocused communication tasks, the designer of the task does not offer prominence to any particular linguistic feature.During the performance of the task learners are not obliged or encouraged to employ particular language structures; the situation should resemble 'natural' communication in which the language used is broadly determined by the content of the task.Focused communication tasks, on the other hand, are examples of functional production practice (Ellis, 2005c, p. 718), that is, they are designed to employ a particular linguistic feature, although not in a way that makes the learner pay more attention to form than to meaning.Therefore, they are also referred to as structure-based communication tasks (Loschky & Bley-Vroman, 1993) or implicit structure-based tasks (Fotos, 2002).While there are some objections and doubts about classroom activities that are planned with specific linguistic features in mind, suggesting that focus on specific grammar forms may not cause any restructuring of the learner's interlanguge (e.g., Long & Robinson, 1998), Lightbown (1998) refutes such arguments claiming that: classroom activities that tend to elicit specific linguistic features need not be awkward and unnatural.They can incorporate the principles of communicative language teaching and task-based instruction, while, at the same time, maximizing the likelihood that learners will have adequate opportunity to be exposed to, use and receive feedback on a wider range of linguistic features.Teachers are not traitors to the cause of communicative language teaching if they plan activities in which they know that learners will almost inevitably need to use specific language features.(p. 195) Communication tasks can become focused either through design or through methodology, and, in their influential paper, Loschky and Bley-Vroman (1993, p. 132) argue that "it is possible to construct tasks which involve grammatical knowledge in various ways, and to varying degrees."As far as the design is concerned, there are three major features focused communication tasks can possess: task-naturalness, task-utility and task-essentialness.Loschky and Bley-Vroman (1993) define the three criteria in the following way: In task-naturalness, a grammatical construction may arise naturally during the performance of a particular task, but the task can often be performed perfectly well, even quite easily, without it.In the case of task-utility, it is possible to complete a task without the structure, but with the structure the task become easier.The most extreme demand a task can place on a structure is essentialness: the task cannot be successfully performed unless the structure is used.(p.132) When the design procedures fail to elicit the targeted language feature, a solution could be found in the methodological choices made by the language teacher.This is because through the teacher's use of brief explicit instructions and corrective feedback, particularly in the form of requests for clarification, the learner may be indirectly encouraged to employ the target language feature.Although the communicative nature of the task is threatened by the focus on form initiated by the teacher, the task will probably remain communicative for the learner who perceives the clarification request as a need to improve the quality of the message.
Tasks designed and conducted according to these suggestions may constitute a valuable instructional option for teaching grammar (Samuda, 2001).
Focused communication tasks can be "a powerful instruction tool in any language classroom where the emphasis is laid not only on acquainting learners with relevant rules but also ensuring that they will be put in the service of successful communication" (Pawlak, 2006, p. 264).Ellis (2003) proposes tasksupported language teaching and claims that focused communication tasks are a useful option in supporting formal instruction with the communicative dimension.Focused communication tasks provide learners with a considerable opportunity to actually use the structures they are taught and convey real messages, which is what most of them may hardly ever do outside of a regular educational context.By means of focused communication tasks learners' implicit knowledge is automatized and, moreover, the transformation of their explicit knowledge into implicit knowledge may be facilitated.Facing the problems of contemporary language teaching, which suffers from lack of time and expects quick results, focused communication tasks may also serve as an indispensable option for review work or remedial teaching (Fotos, 2002;Pawlak, 2004Pawlak, , 2006)).

Issues in Learning and Teaching Past Unreal Conditionals
In their standard form, past conditional sentences are usually constructed with two clauses: the if-clause containing a past perfect verb, accompanied by a perfect modal verb in the main clause, both of which refer to the past.It is possible to reverse the clause order, which does not cause much change in meaning or emphasis.In order to achieve emphasis, one can use inversion techniques in the if-clause.In addition to inversion, the conjunction if can be substituted with other conjunctions or phrases having a similar meaning (Parrot, 2000, p. 237).As far as its meaning is concerned, the third conditional is the past counterfactual conditional describing a situation which is assumed not to have happened.This meaning provokes the contexts in which it is used: excuses, regrets, or expressing relief.Another typical use is presenting alternatives to something that had already happened, often with a tone of blame (Yule, 1998, p. 129-130).It is important to note that the meanings of conditional sentences are also dependant on the kinds of modal verbs used in them.Though it is often assumed that the inclusion of some modal verbs in both the if-clause and the main clauses is incorrect, they are often found in both clauses simultaneously and are part of informal spoken English (Azar, 2002, p. 418).
Since past conditional sentences can express a wide range of functions and their form may indeed result in clauses that are long and difficult for learners to process and remember (Parrot, 2000;Thornbury, 2001), "the so-called third conditional is typically taught at a relatively advanced stage, both because of its syntactic complexity and because it expresses a concept that is itself fairly opaque, that is, hypothetical past time" (Thornbury, 2000, p. 97).Due to the problems inherent in the teaching and learning of conditionals, even advanced students either tend to avoid complex conditional forms, having developed strategies allowing them to communicate effectively without conditional sentences, or they seem to confuse conditional forms which refer to the present and to the past.One way of solving such problems is to apply grammar interpretation tasks (Ellis, 1995) to draw students' attention toward how the form and meaning interact.When it comes to tasks promoting production, problems with meaning and use can also be addressed by providing learners with a clear context or situation typical of the past unreal conditional (Azar, 2002).Fotos (1995Fotos ( , 2002) ) presented explicit structure-based interactive tasks, designed to raise students' consciousness of the correct usage of present and future conditional forms using if.On the basis of positive research results obtained from year-one university students, she suggested that those tasks were a useful communicative activity to improve proficiency and interaction.A study dealing precisely with English past unreal conditionals was conducted by Pawlak (2007).The third conditional was taught by means of two approaches to grammar teaching: planned focus on form and focus on forms.The participants of the quasiexperiment were 102 senior high school students, divided into two experimental groups and a CG.The findings of the study obtained from paper-andpencil tests and dictogloss tasks proved the durable effectiveness of grammar instruction for past unreal conditionals, without a significant predominance of one of the studied approaches over the other.The author concluded that focus on form and focus on forms should be combined in classroom practice rather than viewed as mutually exclusive (Pawlak, 2007, p. 186), pointing particularly to the Polish educational setting, but the claims seem to be valid for the general educational context as well (e.g., Fotos, 2005).

Design of the Study
The main aim of the quasi-experimental study presented in this article was to explore the short-and long-term effects of focused communication tasks on the acquisition of past counterfactual conditionals.The distinction between the two experimental groups was the specific instructional option that was given primary attention for the purpose of the study bearing in mind that isolating the different instructional options is a recommended way of assessing their contribution to learning and drawing conclusions about their pedagogical effectiveness (Ellis, 2005c;Norris & Ortega, 2000).More specifically, the study sought to address the following research questions: 1. Does form-focused instruction facilitate the development of explicit and implicit knowledge of past unreal conditionals?2. What is the effect of employing focused communication tasks on explicit and implicit knowledge of unreal past conditionals in comparison to text-manipulation and text-creation activities? 3. Are the effects of the instructional treatment durable?

Participants
The subjects were 45 year-one full time BA programme students of English at the Teacher Training College, Adam Mickiewicz University, Pozna .Their level of English could be ranked as B2+ according to the Common European Framework of Reference for Languages.The students were divided into three random groups (see Table 1), two of which were experimental and one control.All of them completed the written tests and took part in the recordings: both individually and in pairs.Prior to the testing and treatment procedure, the participants were asked to complete a background questionnaire.Its aim was to obtain information about the students' personal histories and educational backgrounds as far as English was concerned, with a particular focus on English grammar.
Table 1 The design of the study.

Procedure
The instructional treatment took place during regular grammar lessons taught to year-one students on the same days in both experimental groups.The instruction took two weeks, that is, four grammar classes, each of which was 90 min long.At the same time the students in the CG covered other grammar topics and their grammar teacher confirmed that conditionals were not taught.The instructional treatment included the same procedures in the two experimental groups during the first three lessons.For both the FCT group and the CPA group, the treatment started with input-based instruction by exposing the students to a written text containing a number of examples of the third conditional, which was a kind of input enhancement (Doughty & Williams, 1998).All the instances of the target structure were visually enhanced by means of different font type, italics and bolding.Next, the students answered comprehension questions connected with the text (i.e., true/false statements, gap fill, open questions).This part of the lesson aimed at helping the learners make proper form-meaning connections while processing the input by familiarizing them with the context and ensuring their understanding of the meaning of the targeted structures.At this point, the students were not asked to produce the targeted structure.What followed was the analysis of the examples of the feature from the text when the teacher elicited the rule by, for example, concept questions which check meaning and understanding of the structure and guide the learner toward clarifying the rule.The next step was drawing general conclusions concerning the form, meaning and use of the focused grammar area.There was also time allotted to discussing all the subtleties connected with the form under study.Most of these were exemplified in the text provided at the beginning of the lesson.Metalinguistic explanations were supplied as responses to the students' actual queries, connected with the text they read and the exercises they did.The various examples and contexts that the students were provided with encouraged the shift of attention to the form, but it was integrated with meaningful practice in the majority of cases.
The next three classes were entirely devoted to practice, both inputand output-oriented.Comprehension practice, which came first, included such tasks as two-answer alternatives, multiple choice and grammaticality judgement, whereas production practice was organized by means of the following types of tasks: completing with the correct form, transformations and translation.Ellis (1998) recognizes output-based grammar activities of two types: text-manipulation activities, which are highly-controlled, such as paraphrasing or sentence completion, and text-creation activities, in which learners are guided into producing their own sentences using the target structure, for example into creating a story, a dialogue or writing a composition.Ellis recommends text-manipulation followed with text-creation activities, as this order helps learners move from controlled to automatic use of the target structure.Such a procedure was followed in this study.It needs to be mentioned here that during the instructional treatment both explicit and implicit feedback was provided to the students.The fourth treatment session was designed to apply two distinct types of practice in the two experimental groups.One group of subjects spent the whole lesson performing various focused communication tasks, which aimed at eliciting real-life-like communication; the students exchanged information, opinions and experiences on different topics.The tasks were designed in such a way that the use of the targeted structures was highly desirable and advisable to express a particular meaning and reach the com-municative goal.At no time, however, were the students encouraged or advised to employ the structure under study.The students performed the tasks in pairs or small groups and there was also a mingling activity when they worked with various people.
The process of constructing and finding the tasks for the lessons was itself very strenuous and demanding, as it was crucial to meet the three requirements of a focused communication task: task-naturalness, task-utility and task-essentials, proposed by Loschky and Bley-Vroman (1993).As Pawlak (2006) concluded, focused communication tasks promote incidental learning, in contrast to feature-focused activities, which cater for intentional learning with a clear focus on the rule in question.And it was the latter type to which the other experimental group devoted their time during the fourth class.The members of the CPA group continued with various text-manipulation and textcreation activities in which the students were required to use the targeted features.Even a cursory look at contemporary teaching materials and course books indicates that this type of practice is still an integral component of grammar instruction popular among a great number of teachers at different levels.Some activities prepared for the lesson in fact bore resemblance to focused communication tasks, but the main difference was the fact that the subjects were instructed which structure to employ in a given activity.

Instruments of Data Collection and Data Analysis
In accordance with the recommendations of leading SLA researchers (e.g., Ellis, 2002Ellis, , 2006a;;Nassaji & Fotos, 2004;Norris & Ortega, 2000), an attempt was made to establish the levels of the participants' implicit and explicit knowledge at different points of the study in order to provide information about the durability of the instructional gains.Given that "particular outcome measure categories could account for differences observed in the effectiveness of different treatment" (Norris & Ortega, 2000, p. 471), a battery of tests to measure different types of knowledge was designed.The tests constructed for this particular study included: written tests to measure explicit knowledge, and elicited imitation tasks and focused communication tasks to access implicit knowledge.Each of these research instruments was used three times: as a pretest, posttest and delayed posttest.In order to avoid the risk of the practice effect, three versions of every test had been prepared by the researcher.There has been a considerable amount of controversy over the instruments of measuring explicit and implicit knowledge.The research tools employed to estimate the two types of knowledge in this study were chosen, verified and constructed in accordance with Ellis's (2005a) key characteristics of explicit and implicit knowledge.Both the written and the oral tests were analysed employing the following procedures.Having been checked and transcribed, the answers were grouped according to three categories: correct, interlanguage, incorrect.Correct answers, accorded 1 point, were the responses which were appropriate in form, meaning and use for the context provided.Interlanguage forms, given half a point, were such answers which, although appropriate for the context, lacked a fully correct form.Finally, the form was granted 0 points and labelled incorrect if the form, meaning and the use were inappropriate for the context or situation.Apart from accuracy percentages, means and standard deviations were also computed.The data obtained were normally distributed and standard deviations were comparable; therefore analysis of variance (ANOVA) was employed to compare the differences between the three groups on the same test.To assess the significance of the differences in the students' performance on different tests, repeated measures ANOVAs were run.The effect size, aiming to indicate the standardised difference between two means, was established with the help of Cohen's d.
Each of the three versions of the written test consisted of the following six components with instructions supplied in Polish.The first two comprised comprehension tasks, which aimed at tapping the receptive knowledge of the two structures.An untimed grammaticality judgement task included ten sentences, six of which were erroneous.Ellis (2004) argues that given unlimited time, learners have the opportunity to reflect on the sentence, and thus draw on their explicit knowledge.The next task included five sentences of a multiple choice type.The students were asked to choose the correct option out of four provided to complete every sentence.The next four tasks included output-based tasks aimed to tap the students' productive knowledge.The first one was completing ten sentences with the correct form of the third conditional.It was hoped to draw the participants' attention to the required form and thus ensure that they would apply the appropriate rules to complete the sentences.Two types of transformations with five sentences each comprised tasks four and five.As far as task four is concerned, the transformation was quite controlled, as the students were asked to create a new sentence starting with a phrase given on the basis of the context provided.Task five could be done more freely, as there were no restrictions on how to construct a new sentence.The final text-creation task was story completion.Given either some clues or a short story, the students were asked to continue with at least five sentences.
While the written tests were supposed to access the participants' explicit knowledge of unreal past conditionals, it was the implicit knowledge of the two aforementioned structures that was of major interest to the researcher.For the purpose of the study, two different measures of implicit procedural knowledge were employed: elicited imitation tasks and focused com-munication tasks.Elicited imitation tasks have been employed by a number of researchers to measure second language competence.One of the reasons for their popularity in language testing is their reconstructive nature (Munnich, Flynn, & Martohardjono, 1994).It means that, when students hear a sentence, their attention is drawn to meaning and form, and they process the sentence according to the rules which are internalized in their memory.Consequently, learners spontaneously correct the ungrammatical utterances, which allows analysis of their linguistic knowledge.In order to tap the students' implicit knowledge, one needs to meet certain criteria while constructing an elicited imitation task (Ellis, 2006b;Erlam, 2006;Gass & Mackey, 2007;Mackey & Gass, 2005).First and foremost, the reconstructive nature of the elicited imitation task must be ensured by focusing the participants' attention on meaning (Erlam, 2006).The results of Erlam's (2006) study designed and conducted employing such procedures clearly indicate that an elicited imitation task requires the participants to process, rather than repeat, language stimuli.Apart from the relatively easy and straightforward administration and scoring characterizing this test, another asset needs to be mentioned.Unlike many other instruments measuring implicit knowledge, the test allows the targeting of a specific language structure, which undoubtedly adds to its value as far as tapping into students' implicit knowledge is concerned.
According to Ellis (2002), implicit knowledge requires automatic processing, or, in other words, it involves little or no language monitoring.Free language production would then be considered as the most reliable instrument of accessing implicit knowledge.As far as classroom research is concerned, such a quality could be reached by asking students to perform communicative tasks, which ought to resemble real-life situations.For the purpose of the study, three versions of focused communication tasks were constructed both for the third conditional and modals in the past.The same task was administered to all three groups in one test in order not to risk different performance results caused by subtle differences in the tasks.

Results
Written tests measuring explicit knowledge.As evidenced in Figures 1  and 2, and also in Table 2, which depict the mean percentage scores, standard deviations, levels of statistical significance and effect sizes, the instructional treatments turned out to have a similar effect on the students' explicit knowledge of past unreal conditionals in the two experimental groups.The controlled activities employed in the written tests were performed by all three groups on the pretest and the results did not reveal any statistically significant differences between the participants of the study (Figure 1).However, having been subjected to the instructional treatment, the two experimental groups outperformed the CG on the immediate posttest significantly (FCT = 87.81%,CPA = 84.69%,CG = 74.58%)with the effect sizes deemed large (d = .97,d = .79),and also reached statistically significant gains over the CG on the delayed posttest (FCT = 92.81%,CPA = 92.84%,CG = 78.56%,p < .001 in both cases ), which supports the durability of instruction and the carryover of gains for as long as ten weeks.The two groups which had been subjected to two different types of intervention did not reveal any statistically significant differences between the scores they obtained on any test.Looking at the groups and their results separately (Figure 1), the increase in accuracy observed in the two instructed groups was significant from the pretest to the immediate posttest; it amounted to more than 13% in the FCT group (FCT = 87.81%,F = 16.22,p = .001)and more than 7% in the CPA group (CPA = 84.69%,F = 17.96, p = .001).What is particularly interesting, however, is the fact that the groups increased their average scores on the delayed posttest as well (FCT by 5%, CPA by 8%), compared with the immediate posttest results.The group in which the pedagogical intervention included focused communication tasks did not achieve a statistically significant gain, but the delayed posttest score of the other group, instructed by means of textmanipulation and text-creation activities, was significantly higher in comparison with its immediate posttest result.The CG also revealed some improvement on the immediate posttest (CG = 74.58%) in comparison with the pretest (CG = 70.50%);nevertheless the differences became statistically significant as late as on the delayed posttest (CG = 78.56%)compared with the pretest.The findings obtained from the written tests demonstrate that formal instruction did facilitate the development and proceduralisation of explicit knowledge; yet it appears that, for this type of knowledge, the differences in the instructional treatments did not matter significantly.
While the general findings encourage grammar intervention and testify to its effects, which is in accordance with a number of previous studies, some details need further consideration and analysis.Although it is the group instructed by means of focused communication tasks that achieved the best results on the posttest, gaining more than 13% in comparison with the pretest, it is evident from the delayed posttest scores that the group subjected to textmanipulation and text-creation activities did in fact compensate for its poorer gain on the immediate posttest (7%), achieving the same score as the other experimental group.Another fact worth noticing is the change in the SD levels in the three groups.On the pretest, the SD values were quite high, revealing the existence of huge discrepancies in the performance of particular group members.Form-focused instruction appeared to have a diminishing effect on the degree of individual variation in both experimental groups, as SD values decreased remarkably by more than a half on the delayed posttest, whereas the SD level in the CG diminished by 25% only.The changes in the SD values could also be attributed to the overall process of language education at the college, which undoubtedly increased the students' knowledge, developed their language awareness and helped them improve their level of English, which is visible in the SD decrease in the CG.Nevertheless, the SD levels of the two experimental groups seem to be indicative of the beneficial effect of form-focused instruction because the high test scores and the low levels of standard deviations testify to the increase in homogeneity among the participants of the study on the measures of explicit knowledge after both types of pedagogical intervention.

Elicited imitation tasks measuring implicit knowledge.
As visible from the graphical representation in Figure 2 and the detailed description in Table 3, the pretest procedure revealed that the three groups participating in the research project did not differ significantly from each other with reference to the ability to perform the elicited imitation test (CG = 49.25%,FCT = 54.79%,CPA = 50.77%).As indicated by the scores obtained on the immediate posttest (CG = 52.38%,FCT = 83.13%,CPA = 66.67%), the instructional treatment caused significant changes in the two experimental groups; yet it did not generate significant differences between them.In comparison with the CG, the results achieved by the group instructed by means of focused communication tasks was highly significant (p = .003)and the effect size was large (d = 1.11).The group which received instruction containing text-manipulation and textcreation activities approached a statistically significant difference when compared with the CG (p = .06)and the effect size value was much above medium (d = 0.66).A corresponding situation occurred on the delayed posttest, where the two groups differed from the CG considerably, with the FCT group reaching 81.82% and the CPA group 76.82%.The differences were statistically significant both in the case of the FCT (p = .001)and CPA (p = .02)group.

Figure 2
The mean percentage scores for the use of third conditional for the three groups on the elicited imitation test Table 3 The effect of instructional treatment on the use of third conditional on the elicited imitation test The comparison of the results for the particular groups, illustrated in Figure 2, indicates a significant advantage for the instruction involving focused communication tasks.This group made an improvement of more than 28% from the pretest (54.79%) to the immediate posttest (83.13%), with the difference reaching high statistical significance (F = 14.89, p = .002),and the score was only marginally lower on the delayed posttest (81.82%).The growth in the other experimental group was 16% from the pretest (50.77%) to the immediate posttest (66.67%) and another 10% on the delayed posttest (76.82%), with the effect that the difference between the pretest and the delayed posttest was a result that was highly statistically significant, at F = 25.35,p < .001.The CG, the members of which did not receive any instruction in past unreal conditionals, made significant improvement but as late as on the delayed posttest.However, even in this case, as might be expected, the scores were much worse than in the two experimental groups.When it comes to SD values, one can observe a relationship between the results achieved by the experimental groups and their level of variability.There was a comparable decrease in heterogeneity on every subsequent test in the two groups, with the caveat that they started at slightly different levels.The instructional treatment including focused communication tasks affected the students' level of heterogeneity to the greatest extent (a decrease of 9%), when compared with the decrease in SD in the CPA group (a decrease of 4.5%).A finding like this may be reflective of the nature of focused communication tasks which are designed with a view to developing learners' implicit knowledge.The students in the FCT group not only significantly improved their score on the immediate posttest, but they were also definitely more homogeneous in their answers, which seems to be closely related to the employed instructional option.The members of the group instructed by means of various text-manipulation and text-creation activities, which are certainly more controlled in their nature, manifested a much smaller drop of variation on the immediate posttest, which may have been caused by the type of intervention to which they had been subjected.Contrary to the test measuring their explicit knowledge, which they had no problems with, the elicited imitation task seems to have caused them some difficulty, as the immediate posttest score was 66.67% and the answers were more varied than in the FCT group.The CG, having increased their level of variation on the posttest, finally reduced their SD on the delayed posttest, but not to such a great extent as the two experimental groups.Although their elicited imitation test scores were higher on the subsequent tests, it may be connected with their self-study or the practice effect, because the delayed posttest was in fact the third test of this kind that they took.

Focused communication tasks measuring implicit knowledge.
The analysis of the data obtained from pair recordings in which the students performed a focused communication task commenced with calculating the number of obligatory contexts for the use of the third conditional for every student.As evidenced by the graphical illustration in Figure 3 and the data exhibited in Table 4, the three groups participating in the quasi-experiment did not differ significantly in their ability to use the third conditional on the pretest (CG = 68.89%,FCT = 67.05%,CPA = 55%).The comparison of the scores obtained on the immediate posttest following the instructional treatment clearly indicates a significant advantage for grammar intervention (CG = 37.04%, FCT = 89.06%,CPA = 78%).In comparison with the CG which in fact manifested a loss of over 30% (F = 13.70,p = .001),the two experimental groups made some improvement, and the differences between the two groups and the CG were statistically significant (CG-FCT: F = 25.71,p < .001;CG-CPA: F = 9.26, p = .004).The effect size also reached very high values, but the reasons for it may be the poor results of the CG rather than the gains of the experimental students.It may also be of interest that the difference between the performance of the two experimental groups on the immediate posttest approached significance with F = 3.16 and p = .08.The FCT group scored 89.06%, whereas the result achieved by the CPA group was 78%.The delayed posttest results confirmed the variable effects of the treatment between the CPA and the FCT groups as the mean percentage scores differed significantly (FCT = 84.78%,CPA = 61.76%,F = 5.70, p = .02).On the delayed posttest no statistical difference between the CG and the group instructed by means of text-manipulation and text-creation activities was observed (F = 0.08, p = .77).Such findings indicate a significant advantage for focused communication tasks: The score of the FCT group on the delayed posttest was significantly higher (84.78%) than that of the CG (CG = 53.45%,p = .002).Although both experimental groups did worse on the delayed posttest, one may conclude that the treatment that included a number of focused communication tasks was more beneficial to the development of implicit knowledge.To sum up, the FCT group outperformed not only the control but also the CPA group during the test comprising a communication task.When it comes to the analysis of the progress made by the three groups individually (Figure 3 and Table 4), the picture is more complex than in the case of the previous tests.As far as the CG is concerned, the only statistically significant difference in the use of the third conditional was observed on the immediate posttest, on which the students obtained 37.04%, a result that was lower by more than 30% when compared with the pretest (68.89%).One reason for such a poor score may be that the students had noticed that they were tested on something they had not been instructed in, but what seems more likely is that they became tired and bored with the tests and recordings.The potential contribution of the fatigue effect must be taken into account, as the students underwent repetitive, monotonous and time-consuming tests, which might have discouraged them from engagement in the task.On the delayed posttest, the CG was the only one to improve its results from the immediate posttest with a score of 53.45%, but still it failed to reach the pretest score.The comparison of the data for the two experimental groups indicates that the instructional treatment caused positive changes on the immediate posttest, particularly for the group instructed with the help of focused communication tasks, the score for which, in comparison with the pretest, was 22% higher and the gain statistically significant (F = 5.30, p = .04).Taking into consideration the results obtained by the other experimental group (CPA), one could observe that it had the lowest score on the pretest in relation to the other groups (55%), then made a considerable improvement on the immediate posttest (78%), and, what is particularly worrying, experienced a drop of almost 17% on the delayed posttest (61.76%).None of these changes reached statistical significance, and therefore drawing definitive conclusions might be premature.
Although the findings obtained from tasks administered to pairs attest to the beneficial effect of grammar instruction, with the advantage of the treatment containing focused communication tasks, it is necessary to point to some weaknesses which might have influenced the outcomes of the study.One issue might be the choice of the tasks designed to tap students' implicit knowledge of the third conditional in pairs.This test was the first kind of test performed in pairs.Although the students knew their partners and worked with them throughout the study, their results might, to a certain extent, have been affected by the interlocutor.Moreover, despite the researcher's efforts to design three similar tasks for the three tests, they might have generated different output, not only with regard to the content, but also the number of forms provided.All this needs to be taken into account during the analysis of the data coming from this particular measure.This might have been one of the reasons for the disparities in SD values and might also have affected the results of the tests, which were surprising at times.On no previous test were the SD values so diverse.In the CG the disparities did not diminish; on the contrary, SD values rose on the subsequent tests.It may testify to the different levels of the students' knowledge, but also to differences in understanding of the task.When looking at the experimental group instructed by means of textmanipulation and text-creation activities, a similar situation may be observed, with the caveat that the result was slightly lower on the immediate posttest.On a somewhat more optimistic note, the SD values in the second experimental group, in which focused communication tasks were employed, dropped remarkably from the pretest (31.7%) to the immediate posttest (9.5%).Unfortunately, they rose on the delayed posttest (30.6%) to approach the pretest value.This may be indicative of similar levels of knowledge among the students on the immediate posttest, but may also be connected with how they understood the task on the pretest and the delayed posttest.This is particularly important when we consider the fact that the third conditional is often confused with the second conditional, which could cause serious consequences.Although no effort was spared to make the tasks comparable, differences in their levels of difficulty and in their potential for generating samples of language cannot be excluded.These and other important issues concerning the students' explicit and implicit knowledge of past unreal conditionals will be revisited in the discussion which follows.

Discussion
Throughout the study, the researcher's attention was directed mainly to the role of focused communication tasks in the acquisition of past unreal conditionals by advanced learners of English.The instructional treatment including focused communication tasks was compared with another type of form-focused instruction which contained a number of contextualized practice activities, such as text-manipulation and text-creation activities.As indicated by the data accrued in the course of the study, form-focused instruction overall appears to have had a beneficial effect on the development of explicit and implicit knowledge.
As far as the explicit knowledge of past unreal conditionals is concerned, no advantages of one type of instructional treatment over the other were observed, as both produced significant improvements in the results obtained by the two experimental groups in comparison with the CG.The students' ability to deal with comprehension and production tasks tapping explicit knowledge was also measured, and, on the basis of the results of the tests, neither type of instruction can be viewed as privileged, as both experimental groups made similar gains and differed significantly from the CG.When it comes to SD values, which are reflective of the level of the students' variability, the role of the instructional treatment cannot be neglected.Undoubtedly, form-focused instruction caused changes in the learners' explicit knowledge, which is particularly visible on tests measuring the production dimension of this knowledge.The members of the two experimental groups appear to have systematized their explicit knowledge of the two targeted grammatical forms as they produced them accurately in a consistent manner.As far as the behaviour of the members of the CG is concerned, it has to be admitted that the students improved their results in the course of the study despite no formal instruction in the targeted features.This improvement may be attributed to numerous factors, such as the practice effect, exposure-only effect, or maturation, which may have contributed to changes in the results of the groups (Norris & Ortega, 2000, p. 468).This demonstrates that even without the crutch of the instructional treatment, the members of the CG must have become sensitized to the structure after taking numerous tests, which may in fact have encouraged them to study the third conditional on their own.Besides, they must have encountered the feature under study in some meaningful communication contexts, either inside or outside the educational environment.
When it comes to the implicit dimension of the learners' knowledge estimated on the basis of the two tests, one may acknowledge that the two instructional options affected the students' performance in a different way.As the findings undeniably suggest, it was the group instructed by means of focused communication tasks that made the greatest gain in terms of the ability to use past unreal conditionals in their oral performance.It allows a tentative conclusion that the free production component present during the instructional treatment of the third conditional was pertinent to developing the students' implicit knowledge of the target structure.When it comes to the level of individual variation, the impact of the instructional treatment was again considerable; yet it must be noted that during the focused communication task performed in pairs the levels of the students' heterogeneity were quite high and only slightly affected by the intervention.Such a situation could have resulted from the task itself.Despite the researcher's efforts to create tasks which were as similar as possible for the three tests, the students might have understood them differently and might have perceived them as easier or more complex, which could have led to greater variation in their language use than in the case of the written tests measuring explicit knowledge.Another reason for high SD values might have been the context for the activity, that is, a fairly informal conversation with a peer.Moreover, past unreal conditionals themselves are an intricate grammatical feature and are used in specific situations.The students who were concentrating on message conveyance might not have paid their attention to whether they were using the second or the third condi-tional, which might have resulted in higher SD values.It may also testify to the fact that their implicit knowledge regarding this structure was not fully automatized yet.The performance of the CG on the tests measuring implicit knowledge should also be taken into account.While the members of the group increased their results on the elicited imitation test, they performed quite poorly on the focused communication task performed in pairs.When compared with the improvement on the measures of explicit knowledge, it appears that implicit knowledge is not as likely to be developed without any instruction and assistance.The noticing hypothesis (Schmidt, 1990) and the role of conscious attention seem to be very important for the facilitation of linguistic knowledge.Noticing is believed to be of vital importance for the initial registering of new linguistic representations.If the CG was deprived of instruction in past unreal conditionals, then its members might have had problems with the development of implicit knowledge, even though they managed to improve with regard to explicit knowledge of this targeted structure.The comparison of the results achieved by the CG and the high levels of variation they revealed on the tests allows us to conclude that there must have been students who studied past unreal conditionals via self-study, which altogether contributed to improving the scores of the group.

Conclusions and Implications
On the basis of the investigation of the impact of focused communication tasks on the development of explicit and implicit knowledge of past unreal conditionals, it may be concluded that such instruction proved capable of yielding both short-and long-term learning gains.The analysis of the data demonstrated that focused communication tasks were particularly effective for the development of the students' implicit knowledge, as was evidenced in the outcomes of the three measures employed.These tasks developed the learners' awareness of the two forms and helped them increase control over them (Robinson, 2001).Moreover, they seemed to have a profound influence on the learners' cognitive and affective characteristics by fostering their autonomy, self-esteem, self-efficacy, self-confidence and motivation.Therefore, it can be concluded that focused communication tasks are an effective instructional option when it comes to promoting the acquisition of complex grammar structures.It cannot be forgotten, however, that the pedagogical intervention including various text-manipulation and text-creation activities also brought positive results in comparison with the scores achieved by the CG; however, the magnitude of beneficial effects seems to have been smaller here.The differential effects observed in all the three groups may also have been influ-enced by the contribution of other factors, among which individual differences, analysed on the basis of the two questionnaires, seem to have played an important role, which is in accordance with Housen and Pierrard's (2005, p. 9) suggestion that "the learner factor" is one of the crucial aspects to be taken into account when it comes to the effectiveness of instruction for SLA.
It appears reasonable to conclude then that focused communication tasks are an effective, and perhaps even necessary, instructional option.Irrespective of the fact that such tasks are difficult to construct for the teacher and may pose a great deal of problems when it comes to design, they seem to create numerous advantages for learners.They contribute to greater automatization and raise the students' awareness of the target grammar structure.Apart from that, focused communication tasks place the learner in the centre of the learning process and they contribute to fostering learners' autonomy, developing their motivation, and raising self-esteem.It is the learners that are expected to decide what they are going to say or write, and how they are going to do it.Learners can test their hypotheses about the language and, thanks to the communicative goal, they can understand the real value and importance of the quality of their language as a means of communication.All these factors seem to be of vital importance, because, as Robinson and N. Ellis (2008, p. 490) believe, "language is learned from participatory experience of processing input and producing language during interaction in social contexts where individually desired non-linguistic outcomes are goals to be achieved by communicating intentions, concepts and meaning with others."

Figure 3
Figure3The mean percentage scores for the use of third conditional for the three groups on the focused communication task