INTRODUCTION: The topic of collocations in English has been extensively researched in linguistics. Collocation studies are mainly distributional or corpus-based (Sinclair, 1991, 2004; Nesselhauf, 2003, 2005). The focus of these studies is the statistical analysis of the syntagmatic relationship between the components of collocations. In contrast, phraseological studies examine the characteristics of collocations and their lexical environment (Benson et al., 1986; Hausmann, 1989; Lewis, 2000). In applied linguistics, collocations have been extensively researched in vocabulary acquisition studies as an important part of lexical competence since collocational competence contributes to speech efficiency and fluency. Most of the previous studies (Gitsaki, 1999; Wray, 2002; Nesselhauf, 2003, 2005) suggest that second language (L2) and foreign language (FL) learners often experience difficulties with collocation recognition and production, which, in turn, becomes a source of many language errors. Many studies have attempted to identify the reasons for this challenge. Some of the factors that might impact collocational knowledge are intralinguistic while others refer to the learner and the learning environment. Intralinguistic factors mainly refer to collocation-specific features such as frequency of occurrence and co-occurrence, morphosyntactic structure, span, or transparency. Some of the more prominent external factors include characteristics such as the age of onset and cutoff for collocation acquisition, L2/FL input and output, instruction, L2 proficiency, first language (L1) influence, learning strategies, and motivation. To be more successful in their vocabulary acquisition, learners use various vocabulary learning strategies (VLS). These strategies can be task-specific and related to language skills. In this aspect, there are various types of strategies, such as reading strategies or grammar strategies. Collocation learning strategies (CLS) can be defined as specific vocabulary learning strategies for the acquisition of collocations, or more specifically, the actions learners use consciously to assist them in collocation learning. Since collocational competence is the most salient feature of the depth of vocabulary knowledge, this study aims to examine the learners’ self-directed approaches to L2 collocation learning. The use of language learning strategies can be spontaneous or instructed, and it has been associated with higher proficiency in SL/FL contexts (Oxford, 1990; Cohen, 1998; Macaro, 2001; Wong & Nunan, 2001). Strategy instruction has also been linked with better learning outcomes (Plonsky, 2011). This study also examines the effect of motivation on strategy use and collocational competence. Motivation is another variable that has been identified as influential for successful language learning (Gardner, 1985; Ely, 1986; Dörnyei & Csizér, 2002; Csizér & Dörnyei, 2005), and it has been strongly supported by the researchers as a determinant of success in vocabulary acquisition as well (Tseng & Schmitt, 2008; Han, 2014; Zhang et al., 2017). However, few studies have examined the effect of CLS-based instruction on collocational competence, so this study may offer some new insights in this area. Strategy research has embraced self-regulation in language learning as a complementary notion to the existing paradigms of strategies (Gao, 2007; Oxford, 2011). In this view, the qualitative part of the study examined the study logs of the learners of English for specific purposes (ESP) and explored their collocation learning behaviour through the three phases of self-regulation (Zimmerman, 2000). This study is divided into two parts: the first refers to the theoretical framework and the relevant studies related to the research variables. The second part describes the experimental study that was conducted and the qualitative data on the effect of CLS instruction on the development of collocational competence in English for specific purposes (ESP). COLLOCATIONAL COMPETENCE IN ESP: With different approaches in defining and classifying collocations, collocationtargeted research in (applied) linguistics lacks a clear understanding and a comprehensive definition of the concept. It has been established, however, that collocations are combinations of two lexical words (a node and a collocate) that frequently co-occur in language. Some of the collocation-specific characteristics include varying degrees of transparency of meaning, non-compositionality, the strength of association between the components, and span. Vocabulary acquisition is central in teaching and learning ESP. ESP vocabulary consists of specialized technical and semi-technical words. Having this in mind, collocations in ESP have been defined as multi-word expressions containing at least one specialized term which is usually the node of collocation. The most frequent morphosyntactic types of collocations in ESP are noun + verb, adjective + noun, and noun + noun collocations. The main constituents of L2/FL learner's collocational competence are collocation recognition and production. Receptive knowledge mainly involves the knowledge of meaning, while productive knowledge, activated in speaking and writing, involves primarily the knowledge of form (Revier, 2009). So far, it has been established that the productive knowledge of collocations lags behind the receptive knowledge (Laufer & Waldman, 2011; Nizonkiza, 2012). Different traits and factors can be related to collocational knowledge and should be considered when selecting collocations that are taught and tested. One of the most prominent collocation-specific features is the frequency of occurrence. More frequent collocations are acquired and produced more efficiently than the less frequent ones (Durrant & Doherty, 2010; Wolter & Gyllstad, 2013). Some authors have investigated the frequency of nodes (Nizonkiza, 2015), and the results indicate that collocations with more frequent nodes were easier to recognize. Another interesting finding refers to the frequency of co-occurrence or the strength of association between the collocation components. It has been established that collocation processing is affected by their frequency of co-occurrence for native speakers (Ellis et al., 2008). However, L2 speakers' recognition and production are driven by the frequency of occurrence. When it comes to the morphosyntactic structure of collocations, most studies focus on verb + noun or adjective + noun collocations. The first combination has turned out to be more problematic for L2/FL learners of English. Few studies (Gitsaki, 1999; Peters, 2016; Nguyen & Webb, 2016) have included different morphosyntactic types. The results of these studies indicate that there is a stronger association between the components of adjective + noun collocations which helps L2/FL learners to perceive them as a whole. External elements are an important factor in collocation acquisition and those most frequently mentioned include instruction, L2 language proficiency, L1 influence, and psychological-affective factors such as motivation and language learning strategies. Since L2/FL learners often lack authentic L2 input and output, there is a need for systematic instructional support (Wray, 2002). Some researchers support the explicit teaching of collocations (Ying & Hendricks, 2004; Ying & O’Neill, 2009), while others are in favour of a more natural approach in combination with some input interventions such as visual enhancement or input flood treatment (Bishop, 2004; Webb et al., 2013). Many researchers have explored the relationship between L2 proficiency and collocational competence, and they agree that there are collocational errors even at the most advanced stages in both receptive and productive knowledge of collocations (Gitsaki, 1999; Nizonkiza, 2015). Many studies have identified errors in collocation production due to L1 influence, which manifests as approximation, direct translation, or paraphrase (Biskup, 1992; Reder, 2006). These errors depend on the congruency of collocations (whether collocations are similar or different in L1). Except for the language-related external factors, there are also important psychological-affective factors that impact collocational competence. Previous L2 studies show that motivation has been positively correlated to language learning and vocabulary acquisition (Gardner, 1985; Csizér & Dörnyei, 2005; Zhang et al., 2017). The effect of motivation on collocational competence has not been researched so far, but some studies have included language learning strategies in the context of collocation acquisition (Aston, 1997; Liu, 2000; Ying & Hendricks, 2004; Barfield, 2006; Ying, 2010). These studies show that L2/FL users use learning strategies in/when learning/studying collocations. The strategies used are mainly cognitive, metacognitive, or affective, but there are some collocation-specific ones as well (Ying, 2010). The use of strategies has also been linked to metacognition and autonomous learning. Another important issue concerning collocational competence is related to testing. So far, many different test formats have been used, but a very small number of studies have included both receptive and productive knowledge. Also, there is the question of the scope of collocational knowledge. Most of the formats include multiple-choice, fill-inthe-blank, translation, or matching exercises formats, which test only partial collocation knowledge. So far, there has been only one validated test that checks the knowledge of whole collocations, CONTRIX (Revier, 2009). RESEARCH PROBLEMS AND HYPOTHESES: The overall aim of this study was to examine the effect of CLS instruction on the development of collocational competence in ESP. Another aim was to construct a questionnaire for measuring the frequency of CLS use. The study was built around these four research questions (RQ) and hypotheses (H): RQ1. Which CLS are used by ESP learners? H1. The results of this study will show that ESP learners will use a variety of/various strategies for learning collocations, including memory, cognitive, and metacognitive strategies, as well as collocation-specific strategies. RQ2. Will explicit CLS instruction result in a greater frequency of CLS use? H2. The experimental group will demonstrate a greater frequency of CLS use after the explicit CLS instruction. RQ3. Will explicit CLS instruction have a positive effect on collocational competence? H3. After the explicit CLS instruction, the experimental group will achieve better results on the receptive-productive test of collocational competence than the control group. RQ4. Are there any other factors such as motivation that may impact CLS use and collocational competence? H4. More motivated learners in both the experimental and the control group will use CLS more frequently than the less motivated learners. H5. More motivated learners will achieve better results on the receptiveproductive test of collocational competence than the less motivated learners. INSTRUMENTS AND MATERIALS: 1) CLS QUESTIONNAIRE The first phase of this study consisted of developing a CLS questionnaire. The questionnaire contained demographic questions to gather more information about the learners. It also included a set of questions that measured the intensity of motivation (Ely, 1986). The main part of the questionnaire targeted CLS frequency of use. It was designed based on the vocabulary learning strategies questionnaire by Schmitt (1997) and on the student descriptions of the strategies they use while learning FL collocations, obtained from the focus groups prior to the pilot study. When the data was collected, the first version of the questionnaire was tested in the pilot study. The participants resembled the sample that would be used in the main study. The statistical analysis included reliability and factor analyses and the number of items was reduced from 73 to 63. The final version of the CLS questionnaire consisted of 7 factors grouped according to the basic principles of Schmitt’s (1997) organizational framework (discovery versus consolidation strategies): Strategies for the discovery of the collocation meaning Factor 1. Strategies for determining collocation meaning using formal sources Factor 2. Strategies for determining collocation meaning by analysing or guessing Strategies for consolidating the collocation knowledge Factor 3. Organisational strategies Factor 4. Learning collocations in context Factor 5. Strategies based on collocation-specific characteristics Factor 6. Form-focused collocation learning strategies Factor 7. Independent collocation learning strategies using technology and online sources 2) TESTS FOR MEASURING RECEPTIVE AND PRODUCTIVE KNOWLEDGE OF ESP COLLOCATIONS To measure the receptive and productive knowledge of ESP collocations, two tests (pre- and post-intervention measures) were developed. The tests consisted of 21 collocations each, divided into three different formats, seven collocations in each format. The first format tested only receptive knowledge of collocations. It was a combination of a fill-in-the-blank and multiple-choice test, modelled after the COLLMATCH test format (Gyllstad, 2007). The second test, adapted from the CONTRIX test format (Revier, 2009), examined both receptive and productive knowledge of whole collocations. The final part was a translation test, measuring only the productive knowledge of collocations. The items were selected for the tests based on the collocation frequency of occurrence and co-occurrence and morphosyntactic type. An additional selection criterium was to include the collocations containing specialized terminology. 3) CLS INSTRUCTION MODULE Explicit CLS instruction was provided to the experimental group over 4 weeks. The instruction module was designed based on the CALLA strategy-based instruction model (Chamot & O’Malley, 1994). It consisted of six steps: preparation (CLS awareness raising), presentation (modelling of CLS included in instruction), practice (in-class exercises to practice CLS use), strategy evaluation (survey on the usefulness of the CLS), expansion, and reflection. The last two steps of CLS instruction were included in the qualitative part of the research which consisted of a self-regulated collocation learning task using study logs. 4) COLLOCATION LEARNING STUDY LOGS To explore the metacognitive and cognitive collocation learning processes after the CLS instruction the learners in the experimental group were given a task to learn 5 ESP collocations and describe their learning process in their study logs. Study logs were semi-structured and included questions according to the three phases of self-regulated learning (Zimmerman, 2000). In the preparation phase, the learners were supposed to write their learning aims and describe the steps they take before learning collocations. In the performance phase, the learners described the actions they take during the actual studying. In the last phase, their task was to evaluate their learning process and reflect upon it. RESULTS ANALYSIS AND DISCUSSION: The resulting CLS taxonomy shows that CLS encompass both generic language learning strategies (memory, cognitive and metacognitive) and collocation-specific learning strategies. The qualitative study revealed additional CLS that need to be incorporated into the existing taxonomy and tested in future studies. The explicit CLS instruction did not result in a significantly greater frequency of CLS use. The qualitative data showed that a great number of learners were reluctant to adopt the new strategies, because they perceived their current/existing strategies as useful. The explicit CLS instruction resulted in significantly greater frequency of CLS use only in two metacognitive CLS (planning of the learning process and raising awareness about the collocations in the text) and in using collocations in a sentence in order to remember them. The qualitative data showed that a great number of learners was not prone to the adaptation of the new strategies, because they perceived the strategies they currently employ as useful. A significant difference was established in the overall receptive-productive test score between the two groups after the CLS instruction. When different test formats were analysed separately, a statistically significant difference was found only for CONTRIX task scores. The groups did not differ in any other variable (demographic variables, collocation competence pre-test scores, CLS use before instruction) so these differences can be attributed to the effect of CLS instruction. A statistically significant correlation was identified between CLS on Factor 4 (Learning collocations in context) and motivation for both groups before CLS explicit instruction. Interestingly, after the instruction, this correlation was not significant for the experimental group, which may be attributed to the CLS use. The results of the receptive-productive test of collocational competence of the more motivated learners significantly differ from the results of the less motivated ones. The more motivated learners have achieved higher test scores. The scientific contribution of this study lies in the fact that this is the first study to address CLS as a separate construct. The results show that CLS can be described as context-specific task strategies. The instrument for measuring the CLS frequency of use has also been constructed (Appendix 1) and implemented in the ESP learning context. Furthermore, this study also has some methodical implications. It identified the need to include self-regulated processes such as metacognitive control instruction in the formal learning context of ESP. It also recognized CLS as one of the factors that influence collocational competence in ESP. This clearly shows that the activities including CLS use should be implemented in ESP teaching.