Investigation into Learning Strategies and Delexical Verb Use by Thai EFL Learners

Research into delexical structures is not common in Thailand and is in fact a neglected area of investigation despite the fact that knowledge of collocations can improve language competence. This study investigated the learning strategies that caused deviant delexical-verb use by Thai EFL learners. The study was undertaken with two groups of Thai EFL learners of differing English proficiencies who completed a 20-item gap-filling translation test and participated in an interview. The results revealed that learners relied on three major learning strategies native language transfer, synonymy, and overgeneralisation that could lead to uncommon and deviant use of English delexical structures by learners. It was also discovered that both high and low proficiency learners were most familiar with the delexical verb take. The study also revealed that teachers of English should consider the significant role of these strategies when preparing English collocations and lessons on delexical verbs.


INTRODUCTION
One of the most challenging areas in English vocabulary acquisition is collocation, i.e. words that are used together or that are extremely likely to occur together (Sinclair 1991). For example, the expression make a mistake is more likely to be mentioned in comparison with cause a mistake. Other verbs, such as do or create, are not acceptable as collocates of mistake, even though do and create are regarded as synonyms of make. In particular, the structure make a mistake itself can be called a delexical structure, consisting of the delexical verb make and the noun phrase a mistake. Delexical verbs are named accordingly "because of their low lexical content and the fact that their meanings in context are conditioned by the words they co-occur with" (O'Keefe, McCarthy and Carter 2007, p. 38). O'Keefe et al. put delexical verbs in one of the broad categories of basic vocabulary according to their high frequencies in some corpora. Furthermore, O'Keefe et al. also mentioned that delexical verbs are problematic since they occasionally occur with low-frequency items and learners have difficulty understanding the meaning of the whole expression.
In many countries, delexical verbs are popular amongst linguists and researchers (Chi, Wong and Wong 1994, Jukneviçienė 2008, Nesselhauf 2003. However, in Thailand, the study of English delexical verbs is usually not within the interest of Thai researchers, nor is it highlighted in English classrooms. Instead, single words seemed to be the primary focus of Thai EFL teachers and learners. Very few Thai EFL teachers and researchers are aware of the importance of English collocations, which include prefabricated chunks, such as delexical structures, (Boonyasaquan 2006). For example, a research study by Sripicharn (2010) on high-frequency words gave emphasis to only content words like nouns, verbs, adjectives, rather than delexical verbs.
In Thai EFL classrooms, not only are collocations neglected, but delexical structures, which are a subsection of collocation, are rarely mentioned in class. To attain the highest level of effective English language teaching, it is crucial to incorporate collocations, especially useful and common expressions like delexical structures, into teaching and learning materials because "competence in a language involves knowledge about collocation" (Herbst 1996, p. 389). The present study was aimed at examining the learning strategies learners employed and the errors that appeared when learners used English delexical structures, focusing on Thai learners' interlanguage. In addition, the researcher hopes to help raise awareness of English collocation and delexical structure teaching in Thai EFL classes.

COLLOCATIONS
The word collocations was coined by Firth (1957, p. 181) to explain the phenomenon when "collections of a given word are statements of habitual or customary places of that word". McCarthy (1990) made a comparison between collocations and a relationship. That is, some words are 'married' to other words, and sometimes the relationship between words is 'firmly married', as can be seen in a fixed collocation like shrug one's shoulder (Hill 2000). More specifically, shrug can be used only with shoulders, not any other part of the body. Likewise, Finch (2000, p. 152), explained that collocations are related to "the tendency for certain words to occur together".
Using Carter's categorisation of English collocations (1987), Partington (1998) classified collocations into three categories based on degree of restriction: -Fully restricted collocations are a group of words, a member of which can suggest the following or preceding words automatically. An example of this type of collocation is stinking rich. Whenever stinking occurs, rich is normally present with it. -Semi-restricted collocations allow words to co-occur within the phrase, yet still limited in number. For example, the verb harbour is strictly associated with doubts, uncertainty, a grudge, or suspicion, but not any other words. -Unrestricted collocations are able to engage with any lexical items in any circumstances. The adjective fat, for example, can co-occur with numerous collocates as long as it makes sense like fat girls, fat dogs and fat cats.

DELEXICAL VERBS
Delexical verbs are also known by a variety of terms, e.g. light verbs (Butt 2003), delexicalised verbs (Hill 2000), support/ supportive verb (Dura and Gawronska 2008) and lexically empty verbs (Live 1973). The word delexical refers to the weakened lexical meaning in a verb. The shift of meaning from the verb to the following noun is called delexicalisation and the combination of a delexical verb and the following noun is a delexical structure. For example, in the verb phrase take a walk, the verb take is semantically reduced, or it is delexicalised, and the noun phrase a walk receives the core meaning of the whole phrase. Additionally, since delexical verbs are collocational, they are not used alone but followed by a noun or noun phrase. The examples of most common delexical verbs are give, make, have, take, do, be, go, pay, and hold. (Guňková 2011).

SIGNIFICANCE OF CHUNKS
O' Keefe et al. (2007) stressed the significance of chunks and comparing chunks with vocabulary learning and teaching in a conventional fashion. Traditionally, single words were mainly emphasised in vocabulary lessons. They were considered the basic units of meaning which beginners need to pay attention to. It was believed that studying single words was important in the process of building a firm base for language learners. Learners, however, at the higher level of EFL are involved with units consisting of more than one word such as phrases, idioms, collocations or chunks (Kashiha and Chan 2014). Collocations are currently one of the most popular topics amongst language researchers, who often use language corpora as a statistical tool to check how collocations are employed in authentic language. Collocations or chunks like a large number of, at the end of the day, by all means are regarded as 'high-frequency' phrases that appear in daily use as they are ranked in the top 20 multi-word chunks compiled from corpus according to O'Keefe et al. (2007). Therefore, high-frequency chunks should be introduced to learners at any point of time when learning English. The principle of vocabulary learning at the present time is that studying only single words is not enough because chunks are prevalent in our everyday usage of English and therefore, words in chunks are considered significant for learners.
A great number of studies in second language acquisition of collocations, specifically in the area of delexical verbs, have been conducted worldwide. Chi et al. (1994) examined collocational problems amongst ESL learners. The research focused on lexical collocations and the delexicalisation of verbs. The study was corpus-based since the data was drawn from a corpus of 500 word essays by first-year learners at a university in Hong Kong. The researcher investigated the collocational errors in the corpus and the results revealed that get phrases were the most problematic. The findings also portrayed the learning strategies learners used and overgeneralisation seemed to be the strategy that learners relied on the most.

RELATED RESEARCH ON COLLOCATIONS AND DELEXICAL VERBS
There exist a number of relevant studies on collocations and delexical verbs in English. To begin with, Ang, Rahim, Tan, and Salehuddin (2011) conducted a corpus-based study on collocations which focused particularly on types and sources of verb-noun collocational errors. The researchers investigated 130 essays in a sub corpus of a Malaysian learner corpus, EMAS, or the English of Malaysian School Learners. The study was based on interlanguage theory and error analysis framework in data analysis. The findings of the study revealed that of all the seven types of common collocational errors, the most frequently-found was the preposition-related collocational errors. Moreover, of all the three categories of sources of collocational errors, intralingual transfer seemed to be the most noticeable in learners' interlanguage.
With regard to collocations with high-frequency delexical verbs have, do, make, take, and give, Jukneviçienė (2008) investigated the interlanguage of non-native speakers' essays and made a comparison with a corpus of English native speakers. The findings showed that although the awareness of English collocations was raised amongst learners, English collocations were underused. This is probably a result of learners' deficient academic vocabulary knowledge. The study found that direct translation was one of the prominent strategies that the language learners used; and this could have led to the erroneous English collocation forms. Yan (2006) investigated the use of the high-frequency verb do in the English production of Chinese learners. The study was based on the data found in a corpus called the Chinese Learner English Corpus (CLEC). The researcher compared the use of do found in English used by Chinese learners to that of English native speakers in a corpus called Louvain Corpus of Native English Essays (LOCNESS). Using the Contrastive Interlanguage Analysis (CIA) approach, the findings revealed some differences in the use of do between the Chinese English speakers and the native English speakers. The study found that Chinese learners were likely to underuse the verb do in constructing interrogative and negative sentences in English. On the other hand, they overused do with a number of nouns in delexical structures. The three major sources of errors, as shown in the study, were mothertongue interference, intralingual transfer, and overgeneralisation.
However, research on delexical structures in the Thai context is rare. Although collocation is now a popular topic amongst Thai researchers, subtopics on collocations such as delexical structures are apparently not to their interest. A corpus-based study by Sripicharn (2010) seemed to be one of the few studies that covered both areas of collocations and some common delexical verbs. The study was aimed at investigating collocations and patterns of high-frequency words in a small corpus of Thai EFL learners. However, the study mainly focused on content words like nouns, verbs, adjectives, and adverbs, rather than specific delexical verbs. To sum up, there is very little research on delexical verbs in the Thai context at present. The present study was conducted to investigate the delexical-verb use of Thai ESL learners of varying proficiency levels to analyse the learning strategies they used. The study specifically examined the deviant delexical structures in Thai EFl learners' interlanguage. It was hypothesised that certain learning strategies employed by Thai EFL learners will result in deviant delexical structures.

PARTICIPANTS
The participants of the study were 80 Thai EFL learners from various schools in Thailand. The researcher classified them into two proficiency groups: low and high, according to their scores in the Oxford Placement Test. Learners whose scores were lower than 16 were assigned to the low-proficiency group while learners whose scores were higher than 23 belonged to the high-proficiency group. Twenty of the learners whose scores were between 16 and 23 were eliminated to avoid any ambiguity regarding which group they could belong to. The participants were EFL learners between the ages of 14 and 16 years who had been exposed to English as one of the compulsory subjects at school for more than ten years. There were 46 female learners and 34 male learners participating in the study. They were enrolled in a math-science programme and an English-French programme.

RESEARCH INSTRUMENTS
To collect the data, a 20-item gap-filling translation test was used to examine the participants' productive skill. The test was designed to encourage participants to produce English delexical structures according to the test instruction. The researcher administered the test to the two groups of participants and provided the test direction in Thai to facilitate the participants' understanding. The required use of four English delexical verbs was announced clearly and the prohibition of a single-verb use was stated. The test time was set at half an hour and no dictionary or electronic devices were allowed in the test room. Interviews were also conducted with the participants after the test session to verify deviant answers in the tests.

DATA ANALYSIS
The answers from the two tests were quantitatively and qualitatively analysed. To analyse the data quantitatively, the researcher used descriptive statistics, i.e., overall percentages and mean. To tackle deviant delexical structures qualitatively, the researcher consulted The Corpus of Contemporary American English (COCA), which is a 450-million word collection of written and spoken English from a wide range of sources such as fiction, popular magazines, newspapers, and academic, or even between sub-genres such as movie scripts, sports magazines, newspaper editorial, or scientific journals. The Cambridge Advanced Learners' Dictionary (2013) was also used to check the acceptability of combinations found in learner language. Additionally, interviews with the participants were conducted to help confirm the sources of deviant answers. The interviews were aimed at supporting the hypothesis previously formulated concerning the learning strategies; L1 transfer, synonymy, and overgeneralisation.

THREE MAJOR STRATEGIES INVOLVING DELEXICAL-VERB DEVIATIONS
The findings showed that L1 transfer is the source of deviation which approximately half the participants from the high (52%) and low (50%) proficiency groups relied on the most. Synonymy is the second dominant source of deviations which high-proficiency (45%) and low-proficiency (32.5%) learners depended on. The least frequent source of errors is overgeneralization. It is worth noticing that the low-proficiency learners (17.5%) used this strategy much more than the high-proficiency learners (3%). Three major learning strategies played an important role in the production of delexical structures in learner language (Table 1). The strategy that evidently caused most problems was L1 transfer. That is, almost all the high-proficiency learners (38) relied on L1 transfer, leading to deviant L2 use, and all forty low-proficiency learners employed such a strategy. High-proficiency learners (32), who were believed to have more delexical and vocabulary knowledge than the low-proficiency learners (26), relied heavily on the use of synonymy as a strategy. The strategy of overgeneralisation was also found with the lowest frequency in the answers of the high proficiency (2) and low proficiency (14) learners, showing that they overused the delexical verb take because it was probably the most familiar delexical verb amongst Thai learners. The researcher could assume that take was probably the most familiar delexical verb since the majority of the participants in both low and high groups were able to give the targeted translation of delexical structures with take such as take a picture and take a bath.
DEVIATIONS RESULTING FROM NATIVE LANGUAGE (L1) TRANSFER Nesselhauf (2003Nesselhauf ( , 2005 maintained that L1 transfer resulted in problems especially when there was no literal meaning to be compared between L1 and the target language. In the present study, to fill the gap in the translation test, some participants obviously showed the use of L1 transfer by directly translating the given phrase without employing the delexical verb as instructed in the directions. That is, the participants did not only violate the test instruction, but they also relied on their native language to complete the task. (1)*He eats breakfast at the school.
In the example above (1), the target translation was has breakfast. The verb eat was believed to be the direct translation of verb /kin/ in L1 Thai and was not allowed to fill in the gap according to the test instruction. Thai language does not have a literal meaning of the expression have breakfast but a direct translation eat breakfast. Although the example makes good sense and does not violate the rules of Standard English grammar, the single-verb use here is considered a deviant answer since the participant did not produce the expected target delexical structure. The participants whose answers consist of a single verb avoided using the target structure because they were probably not sure about the target answer or did not know the structure at all. In the interview with the participants, some of them admitted that they ignored the test instruction prohibiting the use of single-verbs, and others confirmed that they were not accustomed to the target delexical structure at all.
(2)*Come and see with your eyes for yourself.
In Example 2 the target delexical structure was take a look. The phrase *see with your eyes is not natural for native speakers of English because the phrase itself is redundant. That is, when someone sees something, they must do it with their eyes, not with other organs. The phrase was apparently the direct translation of a Thai common phrase, as shown below: /doo dûey taa khàwng tua ayng/ See with eyes of yourself (3)*Let's walk and play.
Instead of the target expression Let's take a walk, the non-target like sentence in Example 3 was another example of negative L1 transfer. The phrase *walk and play may not correspond to what the learner wanted to mean and also sounds awkward in Standard English. The combination in (3) apparently stemmed from the direct translation of the Thai phrase, as shown below: In Thai / dern lên/ In English walk play DEVIATIONS RESULTING FROM SYNONYMY Synonymy was reported to be a widely-used strategy by Thai EFL learners in a large number of studies (Boonyasaquan 2006, Mongkolchai 2008, Yumanee and Phoocharoensil 2013. Thai EFL learners apparently substitute with synonyms when they learn English and this often leads to collocational errors, or awkward language. In this present study, some of the participants did not produce the target delexical structures with the given delexical verbs. Instead, they filled in the gap with a single-verb or phrase that is synonymous with the targeted structure, violating the test instruction. In addition, the interview with some of the participants revealed that they were not aware of the use of delexical verbs as stated in the test instruction. Moreover, some of them admitted that they did not know the target structures at all. (4)*Meenanuch conspires for Sunday.
In Example 4, the targeted delexical structure was make a plan. Furthermore, the direct translation of the given Thai phrase of /waang phăen/ is to plan. The participants whose answer was conspire were assumed not to interfere with their L1, but they might have relied on the strategy of synonymy. The verbs plan and conspire are synonyms but to conspire refers "to plan secretly with other people to do something bad, illegal, or against someone's wishes" (Cambridge Advanced Learners' Dictionary 2013, p. 322). To be precise, plan and conspire cannot be substituted because of their different facets of meaning where the word plan does not possess any negative connotation like conspire does.
The deviant answer in Example 5 was the expression that does not appear in any consulted major dictionaries and was rarely used by English native speakers. This is because only 7 tokens were found in COCA. Such a combination was not a proper English collocation. The use of the deviant phrase was probably because the participants regarded caution as synonymous to warn since the target delexical structure should be give a warning. In addition, although the verbs caution and warn are synonyms, they are distinguished in terms of register. To be more specific, the word caution refers "to warn someone formally, especially by police" (Cambridge Advanced Learners' Dictionary 2013, p. 232), whereas warn means "to make someone realise a possible danger or problem, especially one in the future" (Cambridge Advanced Learners' Dictionary 2013, p. 1765). Furthermore, in Thailand, signs or notices restraining people from doing something are often written in both Thai and English. They usually have a word like warning or caution as a heading of the message. This probably explains why Thai learners of English view these two words as synonyms and this leads to the misuse of synonymy.
The use of synonymy strategy is also found in Example 6. The verb converse was not the targeted production of a delexical structure; moreover, the use of a single verb was prohibited in the test according to the test instruction. In this case, the use of converse is probably not a direct translation from Thai, so L1 transfer seems not to come into play here. Actually the attempted single verb should be talk. It is probable that the verb converse was in turn considered a synonym of the verb talk, but they are rarely interchangeably used due to their difference in formality. According to the Cambridge Advanced Learners' Dictionary (2013, p. 331), converse is more formal than talk.

DEVIATIONS RESULTING FROM OVERGENERALISATION
The present study showed that learners' answers in the translation tests were also in the form of 'overgeneralisation'and this is similar to the findings in a study on the use of strategies among Malaysian EFL learners (Ang et al. 2011). Even though the strategy was not used as much as L1 transfer and synonymy, Thai EFL learners relied on overgeneralisation in particular test items. The use of such a strategy was similar to previous research studies where overgeneralisation was one of the strategies Thai EFL learners relied on (Phoocharoensil 2011). Thai EFL learners seemed to depend on the strategy when they assumed that some items in the test were similar and the answers were generalisable. Overgeneralisation often leads to awkward and erroneous English combinations.
(9)*She loves to take a hug.
(12)*We should take a talk tomorrow.
(13)*You must take a choice now.
(14)*Parents take a promise to buy their children toys.
As mentioned, the delexical structures with take like take a bath and take a picture were among the most familiar structures for Thai EFL learners. In this study, it was found that some of the participants produced the delexical structures mostly with take in a random manner, as shown in Examples 7 to 14. They sometimes incorrectly matched take with many nouns or noun phrases. Moreover, statistics suggests that the delexical structures with take were the easiest ones for both the low-proficiency learners (M = 57.5) and the high-proficiency learners (M = 79.5), while the delexical structures with make were the most problematic for both the high proficiency (M = 6.5) and low proficiency (M = 59.4) learners.

PEDAGOGICAL IMPLICATIONS
Collocation is believed to be an important area of English learning and the present study was aimed at raising awareness of collocation teaching since collocation is not highlighted much in many English classes in Thailand (Boonyasaquan 2006, Phoocharoensil 2011. The findings of the present study might help related personnel such as English teachers and curriculum developers locate and emphasise the collocational problems, especially those regarding delexical structures, for their learners.
Most deviant uses of delexical structures shown in this study were probably derived from L1 transfer. On several occasions, the participants directly translated the given Thai phrases and produced English translations whose literal meaning did not make sense or sounded unnatural in the target language. English teachers should therefore demonstrate the distinguished elements of literal meaning of Thai delexical structures and their English equivalents.
The findings showed that the delexical verb make is the most difficult for learners to deal with. Thus, expressions of make should be emphasised in class with sufficient examples, preferably from native-speakers' corpus data. The purpose of highlighting the problematic verb in EFL classrooms is to familiarise the learners with the difficult verbs, especially in the form of collocation, to avoid mismatch of the delexical verbs and the nouns following them.

CONCLUSION
In summary, it was clearly portrayed in this study that the three major learning strategies, namely L1 transfer, synonymy, and overgeneralisation, were the prime sources of the delexical-verb deviations. The learners, on the whole, obviously relied on their mother tongue when they were assigned to translate the given Thai phrases. They translated the literal meaning of Thai phrases into awkward English delexical structures because the literal meaning of some expressions in Thai cannot be directly translated into English. In some cases, not only did L1 transfer play a role in the delexical use, but some of the highproficiency learners who were believed to have good command of English also used the strategy of synonymy. The learners did not simply fill in the gap with single verbs that define the literal meaning of the given Thai phrases, but they varied their choices of the synonyms of the single verbs to complete the test. The learners directly translated the given Thai phrases and used their synonyms instead, causing inappropriate collocations in English. Moreover, overgeneralisation was another prominent strategy found in the learners' language. Due to the fact that the delexical verb take was the most familiar amongst Thai EFL learners, as in the expressions take a bath and take a picture, the learners relied too much on the use of take with nouns and noun phrase to form a delexical structure.
Factors, such as the participants' gender, learning background, exposure to international classroom, and social status, may also influence the learners' collocation production. Nevertheless, these factors were not taken into consideration in the present study because the researcher would like to pay particular attention to the production and the strategies used in the test. The present study, moreover, was conducted with only a gap-filling translation task because the researcher planned to analyse only the productive skill of the participants. In addition, only four most common delexical verbs, namely, take, make, give, and have, were investigated in the study. Because of these limitations, it is highly recommended that future studies examine Thai EFL learners' receptive knowledge of delexical structures since the present study focused only on productive skill. In addition, in order to widen researchers' perspectives on how Thai learners generally use delexical structures, more delexical verbs should be included in future work.