L2 Learners’ Receptive and Productive Knowledge of Phrasal Verbs

Phrasal verbs (PVs) are very common in English, indicating their usefulness in everyday settings. However, it was reported that language learners generally have great difficulties in understanding and using this linguistic form. This study investigated Malaysian learners’ receptive and productive knowledge of PVs, and some possible factors which may contribute to their difficulties in understanding and using this language feature. A set of PVs test was developed and distributed to 480 secondary school learners in Malaysia to identify their receptive and productive knowledge with respect to PVs. Real data from the English of Malaysian School Students (EMAS) Corpus was used to examine the actual use of PVs by Malaysian learners. Results of the PVs test indicated that in general, Malaysian learners had an average level of knowledge of PVs. Analysis of the EMAS corpus revealed that they faced a lot of problems in the production of PVs. Some possible factors are highlighted which may contribute to such difficulties.


INTRODUCTION
Grammar has long been regarded as 'the heart of a language', while vocabulary including multi word units (MWUs) like phrasal verbs (PVs) received less attention in language classrooms. However, it is now recognised that it is lexical and not grammatical knowledge that can ensure learners' great proficiency in the target language and lexis is regarded as "the basis of accurate and fluent communication" (Rudzka-Ostyn, 2003, p. 5) and as "an essential part of mastering a second language" (Schmitt, 2008, p. 329).
Despite the importance of vocabulary in language learning, it is generally accepted that L2 vocabulary is a complex process as there are various features or elements of words that learners have to know and understand (receptive knowledge), so that they can be stored and maintained in their mental lexicon and recalled later at the production stage for both written and communication purposes (productive knowledge).

Receptive and Productive Knowledge of Vocabulary
The notion of receptive and productive knowledge is frequently discussed in the area of vocabulary learning (Schmitt, 2010;Webb, 2008). Receptive knowledge entails knowing a lexical item well enough to extract communicative value Published by Australian International Academic Centre PTY.LTD. Copyright (c) the author(s). This is an open access article under CC BY license (https://creativecommons.org/licenses/by/4.0/) http://dx.doi.org/10.7575/aiac.ijels.v.7n.4p.144 from speech or writing, while productive knowledge involves knowing a lexical item well enough to produce it when it is needed to encode communicative content in speech or writing (Schmitt, 2010). The majority of vocabulary is learned receptively and thus it is believed that learners' receptive knowledge seem to be larger than their productive knowledge (Webb, 2008;Schmitt, 2010).
A study conducted by Webb (2008) also indicates that learners who have a larger receptive vocabulary are likely to know more of those words productively than learners who have a smaller receptive vocabulary. Schmitt (2010) argues, receptive and productive knowledge are both important components of overall vocabulary knowledge. In line with his point of view, and as far as PVs are concerned, it is clear that language learners need both receptive and productive knowledge of PVs, not only to indicate their mastery of the target language, but most importantly for them to function well in the real world communication.
a particular meaning (carry out=perform; look into=investigate; go up=increase). In many cases, the use of PVs is believed to be more appropriate and sound more natural in expressing certain ideas (Fletcher, 2005).
Most PVs are metaphorical in meanings and many have multiple meanings which can be literal or non-literal (e.g., take off -remove clothing; leave the ground and fly). It was reported that the 100 most frequent PVs in the British National Corpus (BNC) have 559 potential meaning senses, or an average of 5.6 per PV. This clearly suggests the importance of PVs in language learning and without having good knowledge of PVs and an ability to use them appropriately, it is almost impossible for learners to gain fluency in English and be successful in learning the language.
Despite the importance of PVs in language learning, there is general consensus that they are difficult for L2 learners to master (Littlemore & Low, 2006). Cross-linguistic factors, particularly the influence of learners' L1 and the non-existence of PV structure in learners' L1s may affect learners' understanding of PVs, which may result in the avoidance of PVs (El-Dakhs, 2016;Kamarudin, 2013b). Apart from that, learners' lack of awareness of common collocates, regular patterns and usage, is also reported to lead to deviant or non-standard use of PVs by language learners (Littlemore & Low, 2006;Zarifi & Mukundan, 2014). 'Transitivity' and 'separability' of PVs elements is among other aspects of PVs that can cause further confusion for learners. Learners may also avoid PVs due to their semantic complexity (Houshyar & Talebinezhad, 2012) as most PVs carry multiple meanings which can be literal or idiomatic. It is reported that ESL learners at all levels use less idiomatic PVs (Akbari, 2009).
Taking into account all the factors discussed above particularly the complex nature of the PV itself, learners may require longer time for productive mastery of this language form. Hence, the aim of the present study is to examine the Malaysian language learners' receptive and productive knowledge of this important language feature -phrasal verbs. To be more specific, the study intends to answer the following research questions: 1. What is the level of understanding (receptive knowledge) among the language learners with respect to PVs in general? 2. What is the learners' level of understanding with respect to PVs (receptive knowledge) in relation to their language proficiency level and the types of PVs? 3. What is the learners' actual use (productive knowledge) of PVs and possible factors to the non-standard use of PVs?

METHODOLOGY
A triangulation method was adopted in collecting data for the present study -qualitative and quantitative approach. A set of PVs test adapted from Siyanova and Smith (2007) was used in order to examine learners' understanding of PVs (receptive knowledge). The adapted test was pilot tested and checked for reliability. The final version of the PVs test was also tested for internal consistency and the Cronbach's Alpha value of the 40-item PVs test was 0.806. In general, reliability less than 0.60 is considered poor (Sekaran, 2004). Ideally, the Cronbach's alpha coefficient of a scale should be above 0.7 (DeVellis, 2003;Fraenkel, Wallenn, & Hyun, 2012) to indicate high reliability and ensure good internal consistency.
In the present study, the Cronbach's alpha coefficient of the 40-item PVs test was 0.806, which was higher than 0.7, indicating that the test had good internal consistency reliability, and, thus, acceptable for the study. The test consists of 40 multiple choice questions with respect to common PVs. The test was given to a total of 480 secondary school students throughout Malaysia. Responses were then analysed using the Statistical Package for Social Sciences (SPSS) Version 15, a Software programme used to perform statistical analysis.
To find out learners' actual production (written and verbal) of this language feature, a corpus analysis was carried out. The English of Malaysian Students (EMAS) Corpus is a collection of approximately 472,652 words of Malaysian learners' written and oral texts. The written texts consist of various types of essay including narrative, picture and school based essays, while the spoken data consist of interviews and verbal essays. The size of the written sub-corpora is five times larger than the spoken one, with approximately 402,118 words in the written and 70,515 words in the spoken sub-corpora.

Results of the PVs Test
Results of the PVs will be presented and discussed in order to answer the first question concerning the learners' level of receptive knowledge with respect to PVs in general, and its relation to the learners' language proficiency level and the types of PVs tested.
Results presented in Figure 1 indicated that, from the total score of 40, majority of students scored between 25 and above. Taking a score of 80/100 as a useful cut-off point (to follow the Malaysian school standard), those who obtained 32/40 (80 per cent) and above were grouped as having a good level of understanding, while 24/40 (60 per cent) s was the cut-off point for average level and those who scored below 24/100 were grouped as having a low level Figure 1. Respondents' total score of understanding. Analysis showed that only 175 (36.5 per cent) students scored 32 and above, 289 (70.8 per cent) students scored 24 to 31, and 16 (3.33 per cent) of them scored less than 20 marks. This indicated that, in general, majority of the students (70.8 per cent) had an average level of understanding the common PVs tested.
ANOVA analysis was conducted to find out whether there was any significant difference in the mean score across different levels of language proficiency (low, intermediate, high). Prior to conducting ANOVA, test of normality was carried out and result of the normality test indicated that p>.05 suggesting that the data was normally distributed. Then ANOVA was conducted and results presented in Table 1 showed that p<.05 indicating there were significant differences in the students' mean score and language proficiency level. Table 2 presents results of the post-hoc test which indicated that learners' understanding of the PVs differ across the three language proficiency levels (low, average, high). As expected, high proficiency students showed a better understanding of the PVs tested than the average and low proficiency groups. These results were consistent with those found in the study of Liao and Fukuya (2004). Results of the Pearson's correlation showed that the correlation coefficient value of the two variables was r=0.541, which further confirmed that there was a strong relationship between the proficiency level and students' understanding of PVs. As expected, the result also showed that the correlation was positive, which implied that the higher the level of students' proficiency, the better would be their understanding of PVs. Table 3 shows learners score for both type of PVs (literal or non-literal). Results indicated that the mean percentage of literal PVs was higher (84.09) than the non-literal PVs (71.59), suggesting that non-literal PVs are more difficult for learners to understand. Consequently, they tend to use fewer non-literal PVs (Liao & Fukuya 2004) or avoid using them (Dagut & Laufer 1985). This, however, was not surprising as literal PVs are very transparent in meaning in comparison to the non-literal ones.
Apart from that, results also indicated that high proficiency students showed better understanding of both literal and non-literal PVs (p<0.05). Results of the Pearson coefficient confirmed that there was a relationship between the types of PVs tested and students' proficiency level. While the relationship between literal PVs and proficiency level was at a medium level (r=0.335); the relationship between non-literal PVs and students' proficiency level was relatively big (r=0.553), implying that non-literal PVs are more difficult to understand than the literal ones.
Prior to examining the learners' productive knowledge of PVs, a detailed analysis was conducted to determine the difficulties that the learners faced in the literal and non-literal PVs in the test. Results are presented in Table 4 and Table 5 below. Table 4 above presents the frequency of incorrect answers for literal PVs. Item 8 that tested learners understanding of PV put up was the only item which received more than 50 per cent incorrect responses. This was followed by PVs with AVP 'down' (put down and go down); PVs with AVP 'on' (get on and put on). On the other hand, PVs with AVP 'back' (come back, bring back, get back, go back) received low frequency of errors.
As far as non-literal PVs were concerned, the frequency analysis of non-literal PVs indicated that this type of PVs was difficult to the learners. Results presented in Table 5 shows that five of the non-literal PVs tested (come across, bring up, give up, run into, come out) received more than 50 per cent incorrect responses. On the other hand, only one literal PV tested received more than 50 per cent incorrect responses (put up). This further confirmed that students find non-literal or idiomatic PVs more difficult than the literal ones, which was consistent with the findings reported in previous studies (e.g. Dagut & Laufer, 1985;Hulstijn & Marchena, 1989;Liao & Fukuya, 2004).

Results of the EMAS Corpus Analysis
Analysis of the real data from the EMAS Corpus will be discussed to answer the research question concerning learners' productive knowledge of PVs. Five (5) PVs identified in the PVs test that received high number (>50%) of incorrect answers were further analysed to find out learners' productive knowledge of these PVs. The five selected items were Item 8 -put up; Item 4 -come across; Item 10 -give up; Item 6 -run into, and Item 26 -come out. Following this, the EMAS Corpus was analysed to identify the actual use of these items by the language learners.
PV Put up Item 8 tested the learners' understanding of literal PV put up. A: "Our school will organize a Health and Safety Campaign next week". B: "I know, students will help to _____________ posters all over the school". A. give up B. throw away C. put up D. put on Many students incorrectly answered put on instead of put up suggesting their lack of understanding on the difference between these two PVs in the above context. Examples of put up in the learner corpus indicated that learners seem to restrict the use of this PV by associating it with a small number of object collocates (i.e. balloons and bunting) while other common collocates frequently associated with put up by native speakers (e.g. signs and posters) were not found in the learner corpus.
We were sweap the floor, put up buntings and arranged the chairs. our house. We sweap the floor, put up ballons and clean our house. My and I were staying at home to put up balloons and clean up my aunt Another possible reason why the learners had answered put on instead of put up is perhaps their tendency to decode the individual meaning of put on and assume that posters and signs are usually stuck on something like walls or trees. This perhaps explained why students used put on instead of put up when associating it with posters as in the case of item 8 above. It is important for students to be aware that PVs, including put on and put up have to be learned, understood and produced as a 'unit' rather than a combination of elements in the construction of PVs.
PV Come across Item 4 tested learners' understandings of the non-literal PV come across. A: "I _____________ an interesting book in the library." B: "What book is that?" A: "A book on astronomy, my favourite subject." A. came across B. looked up C. looked after D. came u Result showed that PV looked up was a more popular answer than came across, indicating that students did not really understand the meaning of PVs come across and look up, in the context tested above. A closer analysis of look up in the EMAS corpus suggested that students did understand that look up (check) is commonly associated with nouns that contain 'information' (e.g. books, dictionaries). Below are examples taken from the EMAS corpus to illustrate this.
there are all recent and new. If we look up in there for information we the fix internet for students can look up for information especially now Thus, the choice of looked up in Item 4 above was perhaps motivated by the word 'book' in the dialogue, which influenced them in choosing looked up as the best possible answer. Another possible reason was they do not understand the different contexts in which these common PVs (look up and come across) are commonly used. While look up usually implies deliberate action (we intentionally look for certain information in books, dictionaries, etc.), come across implies unexpected discovery. This is the context that many students may not be aware of, which had resulted in responding incorrectly to Item 4 above.
PV Give up The next item is Item 10, which tested students' understanding of the PV give up. A: "Many people are dying of lung cancer nowadays." B: "Yeah. Smoking could be one of the reasons, I guess." A: "I hope my dad will ____________ smoking." A. put down B. point out C. give up D. throw away Frequency analysis showed that more than half of the respondents (60.8 per cent) answered this item incorrectly.  PV throw away was a more popular answer for Item 10 instead of give up. It was not impossible that the learners selected PV give up as the answer because they combined the individual meaning of give and up. The proper verb 'give' itself has a positive connotation which implies support or encouragement to continue smoking. On the other hand, 'throw' has a negative connotation indicating an action to get rid of something like the habit of smoking. Thus, this might be a possible reason why they selected throw away rather than give up as the answer. Apart from that, the Malay term buang is equivalent to the English word 'throw' which is commonly used when referring to an action to quit smoking as in buang tabiat merokok (to give up smoking). Thus, the learners may have applied their understanding of the Malay term buang (throw), and assumed that buang tabiat merokok was equivalent to *throw away smoking. Tendency to combine the individual meanings of PV elements and translating them directly from Malay to English or vice-versa was rather common among Malaysian learners. This may result in producing non-standard use of PVs hardly found in native speakers' discourse and in real life communication (e.g. *throw away smoking).
Further analysis of give up in the learner corpus indicated that the learners' understanding of give up was commonly associated with issues related to studies (e.g. She advise for me to study hard, preserving and don't give up to other problem); and competitions (e.g. But Nicholas and I did not give up. We wanted to do our best even though we knew that we can't be the winner). Surprisingly, no instance of give up associated with smoking was found in the learner corpus.
PV Run into The next analysis focused on Item 6, which tested the students' understanding of the PV run into. A: "When you think about it, most of your classmates will disappear from your life forever after you graduate." B: "Yeah, but every now and then you will ___________ one of them on the street." A. go over B. get back C. come out D. run into Again, learners' failure to choose run into as the correct answer may due to their tendency to combine the individual meaning of elements in PV constructions rather than treating a PV as one lexical unit. They may find that combining the regular meaning of run and into in the above context, did not make sense at all because a person would not literally run away from long lost friends; instead he/she would go or get near to them. Thus, go over and get back made more sense to them in answering Item 6 above. Closer analysis of the EMAS corpus also indicated that there was no instance of run into found in the learner corpus, indicating the learners' unfamiliarity with this idiomatic PV.
PV Come out Item 26 below tested the students' understanding of PV come out. However, results showed that PV take off was a more popular answer. A: "There's a blood stain on your shirt?" B: "I know. I've washed it many times but the stain just would not ____________." A. come down B. give up C. come out D. take off PV take off is equivalent to the Malay word tanggal, and tanggal can be used in different contexts as shown below. a) tanggal pakaian (take off shirt) b) tanggal kotoran (remove stain/dirt) As take off is equivalent to tanggal, learners may have assumed that PV take off can also be used interchangeably in both contexts. Hence, it was not surprising that many of the respondents had answered take off for Item 26 instead of come out. This again suggested that L1 plays a significant role in the learners' understanding of many common PVs including take off.
Apart from that, sample of concordance lines from the EMAS Corpus shown below indicated that learners' understanding of come out was very restricted to its association with animate subjects (e.g. the girl, the mouse, my brother).
After the girl the girl come out of the river sh she thanks when the cat is away, the mouse comes out and play', so whenever the and combed my hair. After my brother had come out of the bathroom, I saw a Then I had a shower and then I come out of the room. After a week Mere came out of the hospital. All her friends the doctor came out from the emergency room with up to the center of the river. When we came out from the boat and we In fact, there was no instance of come out being associated with inanimate subjects such as dirt/stain found in the learner corpus, suggesting their limited understanding of PV come out in the above sense (Item 26).
In brief, analysis of the learner corpus (EMAS Corpus) clearly showed that apart from the influence of the learners' L1, unfamiliarity with many other senses of PVs which are very common and frequently used in everyday communication, as well as learners' tendency to combine meanings of each element in PV construction may have contributed to the nonstandard use of PVs by Malaysian learners of English.

CONCLUSION
To summarise, results of the PVs test above indicated that, in general, Malaysian school learners have a moderate level of receptive knowledge with respect to PVs. The PVs test conducted also revealed that the learners had problem to understand high frequency PVs commonly used in everyday settings: a useful finding, which was not found in any other research in Malaysia so far. It was also found that language proficiency level played a significant role in the learners' understanding of PVs. Learners with high proficiency language level showed better understanding of PVs in comparison to those with lower level of language proficiency. In addition, the types of PVs tested (literal or non-literal) also had an impact on the learners' receptive knowledge with respect to PVs. Contrary to the literal PVs, the non-literal ones were more difficult for learners to understand.
With regard to the productive knowledge, real data from the learner corpus had revealed a number of problems faced by the learners in the actual production of PVs. Many had a great tendency to treat PVs as two rather than one lexical unit. Apart from that, they were also not adequately exposed to the various core meanings of PVs (literal and non-literal) that are useful to them. Lack of awareness with respect to common collocates of PVs was another factor contributing to the lack of understanding and inappropriate use of this language form. Finally, their L1 was also found to play a significant part in their understanding of PVs.
In short, the above findings provide empirical evidence with respect to the receptive and productive knowledge of PVs amongst Malaysian learners' of English. A number of useful findings have been highlighted, which can inform relevant parties for further actions.