TRANSLATION EXERCISE AIDED BY DATA-DRIVEN LEARNING IN ESP CONTEXT

This paper studies the possibility of using a combination of a simple translation exercise with a small-scale concordance exercise with students enrolled in the English for Tourism course at the Faculty of Economics in Split, Croatia. It explains the rationale behind such a combination of exercises which focus on lexical items and explores how efficient the students were in dealing with the proposed exercises. Although teaching subject-related terminology is an important aspect of teaching an ESP course, the paper presents the case of teaching and practising the semantically light verb take in its frequent subject-related meanings and usages. To cater for this need, a small specialised corpus was used as a starting point and the British National Corpus was queried for verification and comparison. Overall, a simple lexical corpusbased exercise provided useful targeted information concerning the predicted pitfalls in students’ translations. Although only some students actually used this information to improve their initial translations, they were highly successful. The feedback collected in the questionnaires shows that students were generally very satisfied with the experience.


INTRODUCTION
The aims of this paper are to explain the rationale behind combining a simple translation exercise with a small-scale concordance exercise, to analyse how efficient a group of students of Tourism were in critically reconsidering their initial language production (i.e.translations) in the light of the language input provided by a concordance exercise, and to report on students' feedback.The impulse and challenge for combining the two methods was provided by some of their shared characteristics." [T]he process of translation is seen as a slow and laborious one, focused more upon accuracy than fluency" (Cook, 2012: 88).Likewise, the concordance work, focusing on language system rather than fluency and communication, is also sometimes described as time-consuming and laborious (Kabalin Borenić, Marinov, & Mencer Salluzzo, 2013).Furthermore, the two methods hold rather contrasting positions in language teaching theory and practice.Translation is not completely absent from ELT but it was banned from the language teaching theory for around a hundred years and is now being revisited and reconsidered (Cook, 2012).On the other hand, data-driven learning (DDL), i.e. the use of corpus data directly with language students, still suffers from the "lack of Vol.4(2)(2016):  user-friendly applications for general, everyday practice" notwithstanding the researchers who "may be quite keen to underline the potential of their work for teaching and learning" (Tyne, 2012: 114).
Translation is an activity or a strategy naturally employed by both language teachers and learners.Accordingly, despite some possible limitations, its potential should be studied and exploited since it could help language learners gain deeper understanding of what they do when they move between the two languages and thus indicate the areas that might require improvement.Furthermore, potential limitations or pitfalls of translation exercises can be predicted and remedied by carefully designing teaching activities involving translation.For example, Malmkjaer (1998: 6) points out that "translation misleads students into thinking that expressions in two languages correspond one-to-one".It is true that learners may have great difficulties in mastering one-to-many correspondences between L1 and L2 (Stojanovska-Ilievska, 2007;Swan, 1997), as indicated by Heltai's (1996) research of lexical errors in learners' translations at the intermediate level.However, Leonardi (2010) counterbalances the above limitation by indicating the potential of translation exercise in contrasting two languages, which can help learners realise that concepts can be expressed differently.Other frequently mentioned limitations, closely related to this study, are interference and (negative) transfer from L1 that may occur in the process of translation (Malmkjaer, 1998: 6).Interference is present in all situations where a foreign language is being used regardless of the level of proficiency and is therefore to be expected in any language learning situation, including translation exercises.It is a natural product of the encounter of two languages in a "bilingual" mind.Scott and Pavlenko (2008: 217) claim that L1 and translation can help "raise awareness of negative transfer through cross-linguistic comparisons".In addition, cross-linguistic influence theory has offered a more complex view of the positive/negative transfer or interference and has introduced the concepts of preference and avoidance, indicating that L2 users may have preferences for certain types of words or syntactic structures over others (Scott & Pavlenko, 2008: 213).Mastering one-to-many correspondences and L1 transfer are merely two of the many problems of acquiring a foreign language.If they are encountered in a translation exercise teaching or learning can occur.
DDL has been used successfully in training translators, i.e. as a tool facilitating professional translation, one that helps "people who speak different languages communicate in specific situations" (Gile, 2009: 27).Wilkinson (2005) used it with his translation trainees by training them in consulting specialised corpora to inform their translations of tourism brochures.Frankenberg-Garcia (2005) studied her advanced student translators' preference for the resources used in translation (dictionaries, corpora, the Internet, etc.) and found out that although corpora accounted for only 1/10 of all look-ups they proved to be useful in 2/3 of the cases.
Vol. 4(2)(2016):  In the above cited works all students were language students.The present study analyses how students of non-philological studies deal with the proposed exercise since they may have limited knowledge and skills necessary for a linguistic analysis that reading concordance lines usually requires.In the Croatian system of higher education such students are most often enrolled into an ESP course corresponding to their study programme.
Therefore, the emphasis is on the target learners as much as on the employed activity, where DDL is used as a complementary activity to a school translation exercise.Unlike professional translation, school translation exercises "serve mostly as drills for the acquisition of foreign language vocabulary and grammar structures" and are "intended to serve the students themselves" (Gile, 2009: 26).

Outlawing translation
It was the orthodox Grammar Translation Method that gave translation in language teaching a bad name.Introduced towards the end of the 18th century it continued to be used well into the 20th century despite scholars' criticism.It was criticised mostly for focusing exclusively on grammatical accuracy without considering fluency (Cook, 2012).A major shift in language teaching practice occurred at the end of the 19th century with the introduction of the Direct Method which advocated more emphasis on speaking and more inductive teaching.One of its main features is a complete exclusion of students' own language from the classroom.The exclusion of L1 from language classrooms persisted in almost all major methods and approaches initiated since (Cook, 2012).

Towards the laissez faire
The 1970s saw language teaching theory and practice move away from the structural approach.The change was embodied in two new movements: the Natural Approach and Communicative Language Teaching (CLT).They both believed that a successful language learner should focus more upon meaning than form (Cook, 2012).The Natural Approach was developed from the idea that a focus on meaning would activate subconscious cognitive language acquisition processes when stimulated by "comprehensible input" (Krashen, 1985: 2).On the other hand, Communicative Language Teaching aimed at developing communicative competence, Vol.4(2)(2016):  which "included knowing what to say and how to say it appropriately based on the situation, the participants, and their roles and intentions" but this would be learnt informally (Richards, 2006: 9).The two approaches underestimated the role of students' conscious awareness and control of their own learning.
Consequently, the theory of foreign language teaching from the 1970s to the late 1990s was strongly permeated by the idea of learning a foreign language through communication, as opposed to teaching the language forms as had been the case before.Despite its many benefits, the communicative approach, if used alone, may have serious limitations in teaching or learning academic or professional language, since it may fail to provide opportunities to develop accuracy in language use (Hinkel & Fotos, 2002as cited in Machida, 2011).With such a deeply entrenched idea that L1 should be excluded from the classroom and that accuracy should be sacrificed for the sake of fluency, the teaching and learning environment was not favourable for the implementation of the two activities employed in the current study.
It should also be pointed out that modern language classrooms can provide many examples of "friendly teachers and happy students doing interesting, interactive activities" (Scrivener, 2013: 71) but this may not be enough to meet all the (future) needs of students.Furthermore, there is a potential danger of creating a laissez faire atmosphere in language teaching, which may leave the learners with the idea that they can get away with anything as long as the message has been conveyed, or worse still, as long as the speaker believes the message has been conveyed.
As for the fluency-accuracy dichotomy, it cannot be denied that in spontaneous language use it is "the learners' main priority to get their message across with appropriate speed and fluency" but "speed and fluency conflict with accuracy" (Willis, 2005: 8).This, however, does not mean that the exercises focusing on accuracy should be banished but should be provided a place in the teaching and learning practice where there is no feeling that they impede communication.

Contemporary language teaching and learning theory: promising prospects for translation and DDL
An important change in the contemporary language teaching and learning theory was made by re-evaluating the use of L1 in foreign language instruction and it is now given scholarly attention and studied as potentially beneficial (Duff, 1994;Kern, 1994;Liao, 2006;Witte, Harden, & Ramos de Oliveira Harden, 2009).The development of the post-communicative cognitive paradigm has indicated "the necessity for integration of explicit instruction into communicative approaches" (Machida, 2011: 741).Focus on form is seen as complementary and not contrary to communicative tasks and ways of integrating it in language teaching have been Vol.4(2)(2016):  researched (Doughty & Williams, 1998).The status of lexis has been upgraded in language teaching syllabi as equal or even superior to grammar (Lewis, 1997).
Equally, some authors have turned to studying the importance of grammar teaching again in order to enhance proficiency and accuracy (Brown, 1994& Larsen-Freeman, 1991as cited in Machida, 2011).Furthermore, in the Constructivist approach it is the learner rather than the teacher who is given a more active role."It is the learner who interacts with his/her environment and thus gains an understanding of its features and characteristics.The learner constructs his own conceptualisations and finds his own solutions to problems, mastering autonomy and independence" (Thanasoulas, 2001: 1).
Achievements in corpus linguistics have provided a better understanding of how language works (Sinclair, 2004;Willis, 2005), raised a new awareness of the importance of accuracy and restored interest in the language content.A particularly interesting offspring of advanced information technology and the findings of corpus linguistics is the use of corpora and concordancing in the language-learning environment.The idea was introduced as early as 1969 (McEnery & Wilson, 2001) but it was not until the 1980s, when Tim Johns' (1991) concept of data-driven learning was introduced, that interest and further empirical research into the potential of using corpus data with language students was generated.Although research has shown positive results in terms of the applicability of DDL activities, learning outcomes and students' feedback (Boulton, 2007;Curado Fuentes, 2002;Kennedy & Miceli, 2010) it is still not widely used as a regular teaching and learning activity because some consider it to be too timeconsuming or too demanding (Chambers, 2005;Chambers & O'Sullivan, 2004).
Finally, strong advocates of the interventionist teaching believe that "a level of assertive, muscular interventions [are needed to] get more students participating [...] and [to] push and nudge [them] to achieve more" (Scrivener, 2013: 72).This gave rise to the concept of demand-high teaching (Scrivener, 2014), which aims at finding ways of achieving greater depth of tangible engagement and learning.

Aims
As already stated, the main goal of the study is to establish whether ESP students are able to critically reconsider their translations after they have studied the language input provided by a concordance exercise.The main goal, along with the designed procedure, generated a series of specific goals: 1. to predict the L1 interference the students are likely to produce and use it to design a small-scale corpus-informed exercise to provide an immediate response; 2. to determine if there are differences between the receptive (passive) and productive (active) knowledge of the studied items; 3. to find out if a short three-line concordance is enough for noticing the recurrent item; 4. to examine to what extent the students are able/willing to make changes to their initial translations; 5. to investigate whether the success of the exercise is related to previous familiarity with the target item; 6. to find out if there are occurrences of serendipity, i.e. incidental findings besides those concerning the target item(s); 7. to identify the potential of the actual exercise spilling over into new activities; 8. to get feedback on students' overall satisfaction with the exercise(s).

Participants
A total of 55 first-year undergraduate students of Tourism at the Faculty of Economics, University of Split, Croatia took part in this study in the first semester of the academic year 2013/2014.English for Tourism 1, 2, and 3 are obligatory courses in the first three semesters of their study programme and the course level is adjusted to suit the expected students' proficiency upon passing the school leaving exam (B2 level of Common European Framework for Languages).The contact hours are organised as one hour of lectures (the lecturer delivers an excathedra lesson/lecture with minimum student participation) and two hours of class time (students are required to participate in a variety of language class activities) per week, which makes a total of 15 hours of lectures and 30 hours of class time per semester.

The task
Bearing in mind the importance and complexity of teaching and learning vocabulary, the task was designed with the following possible contributions in mind: Vol. 4(2)(2016): 225-250 5. instilling the principles of the constructivist approach by letting students build on their prior knowledge and exploiting the opportunity to acquire new knowledge from authentic experience.
The task focuses on teaching/practising vocabulary due to the importance and complexity of teaching and learning vocabulary but also because in corpus-derived material it is easier for learners to notice and study lexical information (Gaskell & Cobb, 2004) and concordance work caters particularly well for the depth of vocabulary acquisition (Cobb, 1999).
The language items addressed in the task are the following three usages of the verb take: (i) in the sense of to last, (ii) as part of the phrase take to (the streets) in the sense of to make for/flee to, and (iii) as part of the phrase take place.The items were selected based on their frequency in the small specialised corpus and the lecturer's assumption that these would provide an opportunity for a more indepth vocabulary acquisition of a familiar item.
The emphasis is thus on a big theme despite the criticism that DDL research deals with "minute details of the phraseology of particular words, and may be difficult to reconcile with the big themes of language teaching, such as tenses or articles" (Hunston, 2002: 184 as cited in Boulton, 2007).The focus of this paper is neither on tenses nor on articles but on the verb to take.As a semantically light verb, take obtains a large range of meanings depending on the items it co-occurs with and can, therefore, be particularly problematic for learners of English.Lexicographers constantly struggle to classify its many senses in the most userfriendly manner (e.g.there are 26 senses listed in MacMillan Dictionary, and 64 senses in Collins English Dictionary, excluding many fixed phrases).The complexity of the verb to take allows us to hypothesise that many of its frequent senses or usages have remained unknown or untapped in students' language expression.
Finally, although a great deal of ESP vocabulary learning may focus on subject-related terminology it is also necessary to study relatively well known general words but in frequent subject-related meanings/usages.
The task required no previous training.It was assigned to students as a paperbased exercise and it consists of four parts: The three Croatian sentences were chosen from the Internet and they exemplified the usage of the lexical items targeted by the selected concordance lines.Students were first asked to translate the sentences (from Croatian into English).Then, they were given the three sets of concordances to study.Although Vol. 4(2)(2016): 225-250 general language items were being studied, all concordances were generated from the Mediterranean Corpus in order to provide the context of tourism for the students of Tourism who participated in the research.The concordance material was given to students as paper-based, three-line concordance sets, each illustrating one target language item.Despite the fact that such a small sample of language may be denied the status of a DDL exercise, even Johns' (1991) multiple contexts for an item frequently featured as few as three lines (Boulton, 2010).
Finally, the students were asked to reconsider their initial translations by using newly discovered information from the concordance lines.The students' initial translations were not collected before the end of the whole process, i.e. they had access to them at all times.The emphasis was on improving, reconsidering, polishing rather than correcting because it could not be assumed that an error had been made.In other words, the study wants to determine whether students are able to critically reconsider their translations after they have studied the language input provided by a concordance exercise.

Resources
The concordance exercises were generated from a small, specialised corpus of 450,000 tokens, called the Mediterranean Corpus.It is a compilation of authentic written texts of tour guides of the Mediterranean countries, which makes it a oneregister (tourism) and one-genre (tour guides) corpus.Originally, it was compiled in a fashion of an ad hoc corpus as a source of corpus-informed exercises in a project realised with students of Tourism (Marinov, 2013) and is now being used regularly for teaching purposes to address particular language issues in the same ESP context.The British National Corpus (BNC) was searched using Sketch Engine for comparison, verification, and as a source of information and ideas for follow-up exercises.The Internet provided authentic L1 sentences that the students were required to translate.
The task was completed in class and the time allocated for the task was as follows: 10 minutes for translating the three sentences, 20 minutes to study and translate the sentences highlighted in the concordance (from English into Croatian), and 10 minutes to make improvements to the initial translation.The last 10 minutes were devoted to completing a feedback questionnaire.Students were allowed to use dictionaries.
Students' feedback was collected using a purpose-made anonymous questionnaire. 1 Although anonymous, it was coded, i.e. the students were asked to use the same code on the questionnaires and on the task sheet for the research needs.The questionnaire consisted of 18 questions: 14 five-point Likert scale Vol.4(2)(2016): 225-250 questions, 2 multiple choice questions and 2 open questions where students were asked to express their opinion of the exercise.
The first two Likert scale questions asked about whether the students had difficulties translating the sentences in the first attempt and whether they were satisfied with the achievement.
The second group of questions (ten questions) asked more specific questions about difficulties in translating (whether finding a Croatian equivalent proved to be a problem in the case of translating the concordance examples and whether unknown English vocabulary represented a difficulty) as well as about students' perceived passive or active knowledge of the studied items.
The third group of questions was aimed at eliciting students' attitudes towards altering the initial translation.Students were asked whether they made any changes to the translation in the second attempt; whether the concordance exercise they did was helpful in making the change; whether the lexis emphasised in the concordance exercise was what they needed to increase the quality of translation; whether they lacked other expressions, apart from the targeted one, to improve their initial translations.
Finally, the last group of two open questions asked the students about what advantages or particular benefits of this exercise they could see and to whom they would recommend this type of exercise.

ANALYSIS AND RESULTS
The analysis starts by explaining the response rate to the activity, followed by the analysis of students' responses to the given task, separately for each item observed.This is followed by the results of the questionnaire which indicate the students' perceptions about the difficulties of the task, and about their passive and active knowledge of the studied items.Finally, the students' attitude towards the overall task is analysed.

Response to the activity
Slightly less than half of the students (45.5%) made an attempt at improving their initial translations in light of their findings in the concordance.The remaining 30 students (54.5%) made no changes.
A total of 20 students (80%) who made improvements agreed that the concordance translation exercise helped them change their initial translations.A slightly smaller number (64%) were of the opinion that the concordance translation exercise provided the very lexis they needed, because they had also lacked other expressions for a better translation.

Sentence One
Students' translations of the first sentence were expected to show mother tongue interference or rather preference for the verb last and avoidance of take to indicate duration by native Croatian speakers.Concordance 1 was then supposed to present this neglected item as a useful alternative.Table 3 is an extract of students' responses.The target items are in bold.Changes not related to the target items are underlined.
Concordance 1. Three-line concordance illustrating the usage of to take in the sense of to last; Source: Mediterranean Corpus Twenty three attempts at improvement were made and 22 times it involved the target language item.Most often (20 times) the crossing/trip/journey was said to last, which was expected because take used in this sense is neglected in the production of many Croatian speakers.
Besides the changes of the target item there was a small number of other changes influenced by the broader context.For example, five times the boat trip replaced ferry drive or ferry ride or was inserted where students had been unable to do more than indicate their uncertainty as to how to translate this part of the sentence.Four times at least was also borrowed along with the target item take from the concordance although it was not needed.Likewise, there were five inappropriate applications of the target items, i.e. the verb take was overused in contexts where it did not fit (marked with * in the table ).
Students' translations revealed some other easily predictable L2 solutions influenced by L1, e.g. to go instead of to leave.The verb leave in the sense of depart, used with other means of transport as well, can therefore be a starting point for a new exercise.
Living in a country with more than 1,000 islands, the word ferry is likely to prove useful and important to Croatian students studying tourism.Sketch Engine's (Kilgarriff et al., 2014) Word Sketch function (Figure 1) reveals some other items worth exploring: how ferry is used (i) with frequent verbs such as run, operate, take and (ii) as a modifier of the noun crossing which in the minds of Croatian native speakers is related (almost) exclusively to crossing a street.The number of BNC examples is rather modest but would suffice for an exercise.
A farries from Split to Supetar go every two hours.The farry ride Split-Supetar takes one hour.
From Split to Supetar ferries goes about every two hours.Driving with ferrie Split-Supetar is 1 hour.
The boat trip from Split to Supetar goes about every two hours and takes 1 hour.
Table 1.Initial translation and altered translations of the first sentence specialised corpora in teaching ESP where the occurrence of subject-specific terminology can be higher than in a much larger more general corpus.
Concordance 2. Extract of the concordance output for the noun ferry in the Mediterranean Corpus; sorted alphabetically to the right

Sentence Two
For the second practised item (take to the streets), 12 students were successful at making a correct change and substituted their initial choice with the one occurring in the concordance exercise (extract in Table 4).Another 6 students understood it was about take but did not manage to grasp the usage successfully and failed to recognise the complex colligational and collocational relationship of the four elements and therefore failed to make the desired change (marked with * in the table ).Only one student used an item from the broader context and successfully replaced year festival with annual festival.
L1 sentence: Stotine tisuća ljudi izlaze na ulice Rio de Janeira za svoj godišnji festival.(Hundreds of thousands of people take to the streets of Rio for their annual festival.3) initial translation students' altered translations Hundred of thousands of people come out on the street of Rio de Janeiro for their year festival.
Hundred of thousands of people come out on the street of Rio de Janeiro for their annual festival.
Hundred of people goes on the streets of Rio de Janeiro for their year festival.
Hundred of people take to the streets of Rio de Janeiro for their year festival.

Vol. 4(2)(2016): 225-250
Hundred thousands of people go out on streets of Rio de Janeiro for their this year festival.
Hundred thousands of people take streets* of Rio de Janeiro for their this year festival.
Every year hundreds of thousands people are on streets of Rio de Janeiro because of the festival.
Every year hundreds of thousands people take to streets of Rio de Janeiro because of the festival.
The hundreds of thousands of people are going on the streets of Rio de Janeiro for their yearly festival.
The hundreds of thousands of people take place on the streets* of Rio de Janeiro for their yearly festival.
Hundreds of thousands of people go out on the street of Rio de Janeiro for the annual festival.
Hundreds of thousands took the street* of Rio de Janeiro for the annual festival.
Hundred thousand people go on the streets of Rio de Janeiro for their festival of the year.
Hundred thousand people take to the streets of Rio de Janeiro for their festival of the year.
Hundred thousand of people are going out on streets of Rio de Janeiro during their ?festival.
Hundred thousand of people take to streets of Rio de Janeiro during their ?festival.
Hundred thousands of people go out on the streets of Rio de Janeiro for their year festival.
Hundred thousands of people take the streets* of Rio de Janeiro for their year festival.
Hundreds of thousands people go out on the streets of Rio for their annual festival.
Hundreds of thousands people take to the streets of Rio for their annual festival.
More than hundered thousand people go on streets* of Rio de Janeiro for own jubilee festival.
More than hundered thousand people takes streets* of Rio de Janeiro for own jubilee festival.
Table 2. Initial translation and altered translations of the second sentence Concordance 3 was offered as a source of information regarding the second item in question.
Concordance 3. Three-line concordance illustrating the usage of take to the streets; Source: Mediterranean Corpus In their translations, students mostly opted for go out on the streets or go to the streets, prompting us to check both items in a bigger corpus than the Mediterranean Corpus.BNC search resulted in 12 hits for go out on the street(s) and a further 9 for go on the street(s).Apart from low frequency a more careful look into the extended context of the given lines indicates contexts different from the one in the translation exercise sentence.These findings possibly lend themselves for an exercise in studying the semantic prosody as an important component of a unit of meaning.Semantic prosody was used by Louw in 1993 to indicate that words "tend to occur in particular environments, in a way that their

Students' perceptions
When asked about their satisfaction with their first attempt at translation as many as 76.4% students claimed they did not encounter any particular problems with the translation.In addition to the assumed problem caused by unknown English vocabulary almost half of the respondents also reported problems with finding good Croatian equivalents for some items (49.1%).On the other hand, 32.7% reported having problems with new English vocabulary, 45.5% had no such problems, and 21.8% were undecided.
Vol. 4(2)(2016):  Expecting to find some discrepancies between the receptive and the productive knowledge of the suggested items, the researcher asked the students (i) if they were familiar with the meaning of the target item(s) and (ii) if they used it/them regularly when speaking English.Only 8.1% were not familiar with the usage of take in the sense of to last and as many as three quarters were familiar with it.The remaining 16.4% were undecided.Also, nearly half of the students claimed they used this item regularly while 12.7% admitted to not using it.The remaining 38.2% of the participants were undecided.This is far from what they had shown in their translations, where take was used only once.
A total of 30.9% of students were not familiar with the phrase take to the streets, many were not sure (23.6%) while the remaining 45.5% claimed they were familiar with its meaning.As many as 41.8% said they did not use this item, 32.7% were undecided and 25.5% of the participants claimed to use it.Low level of productive use of this sense was further emphasised by the translations where take was not used even once in this sense.
Finally, a quarter of the students were not familiar with the usage of take place, 16.7% were undecided while almost 60% were familiar with it.On the productive side, there was an equal number of students who claimed they used it (34.6%)and those who did not use it regularly (34.5%) and the rest of the students were undecided (30.9%).The item was used twice in their initial translations.
Familiarity with … Mean take = last 3.9455 take (to the streets) = make for … 3.1636 take place 3.4444 Table 4. Extent to which students are familiar with the usage of the three suggested language items Usage of … Mean take = last 3.4182 take (to the streets) = make for … 2.8000 take place 3.0364 Table 5. Extent to which students use the three suggested language items actively While the above percentages confirm the assumption that the selected language items require more effort to move them from merely passive (receptive) knowledge into the learners' active (productive), Tables 4 and 5 summarise the means gained for these answers.The means show the relative extent to which the students find the items in question familiar (Table 4), i.e. the extent to which they Vol.4(2)(2016): 225-250 use it (Table 5).It is interesting to note that for the "familiarity" question none of the items scored less than 3.This means that, overall, students are fairly familiar with the item but also that they are not fully confident of this fact.The means indicating the students' perception of usage clearly show that students are aware of the fact that these items are not yet fully activated in their language production.The lowest score for the active use is obtained for the item take to the streets.As shown by the above analysis of the students' translations these perceptions are confirmed and even sharpened showing that students' perceived usage may be even lower than they indicated.

Students' attitude towards the task
The majority of students stated that the exercise provided an opportunity for extending vocabulary and facilitating vocabulary acquisition.A number of students were more specific about their experience.A total of 16 students would recommend these exercises to "everyone".Others mentioned those who have problems with a lack of vocabulary, several limited it to university and high school students and one suggested all language teachers should use it in their classrooms.Finally, one student recommended it to those "who always have a problem finding the right word to express what they have to say" and another to those "who like doing things and learning on their own, thinking and helping themselves".

DISCUSSION
The findings of the questionnaire confirm that students may have problems finding the right expression not only in a foreign language but also in their L1.In other words, they may be able to communicate in a foreign language but they may lack words to express the exact meaning, which then has an adverse effect on their fluency.Likewise, it has been shown, quite predictably, that students' receptive skills are stronger than their productive skills, which means that translation exercises could be beneficial for developing both students' L1 and L2.
In terms of achievement, the most effective exercise was the first one, in which the students made 22 successful attempts to alter the initial translations by applying the target item presented by the concordance.In the questionnaire, the students claimed to be most familiar with this usage of the verb take and, correspondingly, they believed they used it actively as well.This claim was, however, not supported by the evidence of their first attempt at translation, in which it was used only once.Students did, however, recognise take as a suitable substitution of the verb last and were willing to introduce it in their translations instead of it.
Although students claimed they were less familiar with the use of take in take to the streets than with take place it was shown they made more changes with the less familiar item.However, several changes were unsuccessful, which probably means that it was more difficult to grasp the meaning and the usage of this item and successfully apply it where needed.In the third exercise, 12 students changed their initial organise for the concordance target item take place.It can be assumed that students had no problems recognising the use of take place since they were quite confident about its meaning although they admitted they did not use it very often.Rather than encountering problems in identifying the target item, students might have been reluctant to change anything because they were satisfied with their first translation (to organise) which in meaning and form closely corresponds to the Croatian verb organizirati.This exercise performed the useful function of showing that there are other options available, thus potentially adding to the range of students' vocabulary.
Furthermore, many students' decisions not to modify their initial texts may be a reflection of the manner in which they most often receive feedback in terms of error correction.That is, they rely on the authority of the teacher to correct them.Accordingly, they may find themselves at a loss when required to look critically at their work and self-correct or rewrite it (Marinov, 2011).It should be one of the aims of language exercises to help raise students' awareness of multiple possibilities in a language without one choice necessarily being proclaimed an error and the other one(s) its desirable replacement(s).Accordingly, the exercises discussed above enable students to creatively engage with the language they are accustomed to using and to explore alternatives in the given context.
Vol. 4(2)(2016):  Apart from the changes to the target items, students were able to make further changes as a result of their exposure to corpus data.This is evidence of the possibility of serendipity in corpus consultation, even in a carefully guided exercise and a small-scale concordance task.On the down side, some of the changes made were inappropriate.
In light of the above findings it can be said that the proposed exercise encouraged the students to construct knowledge themselves by matching new against given information and by establishing meaningful connections, thus fostering autonomy and independence.At this level of language proficiency learners are able to recognise an alternative solution when they hear/see it but the task has to be kept simple and straightforward.There is no point in insisting on complex tasks with novice and/or less proficient language users or students whose primary aim is not the study of language.As with other types of classroom activities, "DDL activities can be plotted on a cline of learner autonomy, ranging from teacher-led and relatively closed concordance-based activities to entirely learner-centred corpus browsing projects" (Mukherjee, 2006).In this case, the concordance exercises were simple enough to handle and led students to the target items providing the opportunity to directly apply the newly-made discoveries.This way a learner-centred approach was achieved, striking a balance between the traditional ex-cathedra language presentation and having the learner take on the role of language researcher by letting them research language, albeit on a nonthreatening scale.The task also proves to be suitable for the demand-high teaching, which claims that we should more often get to grips with "the language the students produce, wrestling with sentences and errors, helping them to explore their way forward [by helping them to] become more alert to learning [and] to see what they are doing and discover other ways to do it or do it better" (Scrivener, 2013: 72-73).As remarked by Cook (2012: 101), "translation might sometimes be useful to learners in formulating what they have to say or write, precisely because it slows them down, allows them to consider carefully what they are saying...".DDL exercises may also produce the same desired effect of slowing down and allowing time for intense practice and focus on details that are otherwise neglected in the rush to get the message across quickly.

CONCLUSION
The main aim of the study was to find out whether students are able to critically reconsider their initial language production (translations) in light of the language input provided by a short concordance.Out of the 55 students who took part in the study 25 were capable and/or willing to complete the exercise thoroughly, i.e. all three items.The remaining 30 completed the first two parts (L1 to L2 translation and the consultation of concordances) but did not make any changes to their translations using the concordance findings.At this point no conclusions can be Vol.4(2)(2016): 225-250 drawn as to whether they were not able or they were not willing to make changes, because the students were not obliged to change anything.In the 25 completed assignments students were able to identify the target item in the three short sets of concordances.As for the specific goals of the study the findings showed that: 1.An experienced teacher sharing the students' mother tongue can predict the L1 interference students are likely to experience and design an appropriate smallscale corpus-informed exercise to provide an immediate response to this.A simple, lexical, corpus-based exercise, such as the one(s) exemplified here, should not be beyond the capabilities of anyone involved in language teaching today.Allowing that the preparation of such exercises is time-consuming and teachers have little time to invest in it, we propose producing a neatly structured and easily accessible database of corpus exercises.Teachers like using simple, ready-made, innovative materials and it is likely that such an initiative may eventually attract the teaching community to contribute as well.2. The students admitted there was a difference between their receptive (passive) and productive (active) knowledge of the three studied/practised items.This was further confirmed by their actual language production.3. The three-line concordance format appeared sufficient for presenting and noticing the target items.4. The analysis showed that students were highly efficient in applying changes: 88% attempts in the first sentence (all successful), 72% attempts in the second sentence (48% successful), and 48% in the third sentence (all successful).5. Previous familiarity with the items is not directly related to the success of the exercise.There are other factors that may influence the outcome.The assumption is that more difficult items are not an insurmountable obstacle but may present a welcome challenge (e.g.sentence two).Likewise, some easier items may be recognised but not applied if the initial translation is clearly acceptable (e.g.sentence three).6.Some serendipitous usage/learning was noted although the application was not always correct.This is worth noting because it indicates students' ability to recognise (i) recurrent patterns, and (ii) (better) alternatives.7. The exercise triggered new ideas for potential follow-up activities based on the learner corpus created in the process (students' translations) and the evidence of a larger corpus (the Mediterranean Corpus or BNC).8. Students' overall satisfaction with the task as reported by using a questionnaire shows that they were highly satisfied with the experience.Although this is not equal to measuring the actual positive learning outcomes, such positive evaluation is of course highly desirable and is noteworthy."[I]ndicators of learner engagement or achievement" (interest in the work, completion of the exercise, etc.) are "a simple but effective and perfectly normal means of seeing how well a given activity is working" (Tyne, 2012: 123).
1. translating three sentences from Croatian into English; 2. studying three short sets of concordances, each illustrative of one lexical item; 3. translating the highlighted sentence in the concordance from L2 into L1 to retain focus on the practised items; 4. reconsidering the initial translations (based on the study of concordance examples).

Figure 1 .
Figure 1.Extract from the Word Sketch of the noun ferry Concordance 4 was provided for students' reference regarding the third itemtake place.Vol.4(2)(2016): 225-250 Concordance 4. Three-line concordance illustrating the usage of take place; Source: Mediterranean Corpus In addition to the target item (synonym of organise as a collocate of concert), the translations reveal several other points that may be tackled by a corpus-informed exercise.The phrase in/on the open space(s) is much more frequent in this small learner corpus (the students' translated sentences) than it is in the BNC where it is featured only five times (Concordance 5).Although this is a very small sample of language it clearly illustrates the usage of the phrase and also indicates the potential difference between using the singular or plural of the noun space.Providing extended contexts of these concordances could form the basis for establishing contexts in which the phrases are used.Likewise, other exercises derived from the concordance of the node in the open sorted alphabetically to the right are recommended.A corpus-based comparison of in the open and outdoor(s), another frequently used lexeme in students' translations, might also bring about interesting and revealing findings.Concordance 5.In the open space(s) as featured in the BNC ):L1 sentences: Iz Splita za Supetar idu trajekti otprilike svaka dva sata.Vožnja trajektom Split Supetar traje 1 sat.(The ferries from Split to Supetar leave every two hours.2 )

Table 3 .
Initial translation and altered translations of the third sentence Their observations and comments on what they found most useful about the exercise are summarised in the following list: