Collocation research mainly in terms of L1 influence

Chapter 3. Literature review (2):

Empirical research on collocation

3.1. Introduction

In comparison with the collocation studies discussed in the previous chapter, empirical research on collocation from the perspective of L2 acquisition has been extensively conducted for the last two decades. This chapter briefly reviews the main empirical research on collocation to date.

collocations. Her subjects were instructed to supply the English translation equivalents of lexical collocations in Polish and German separately. There were two main findings she discovered in her experiment. One finding was that the quantitative and qualitative results had something to do with different teaching policies in Poland and Germany. The Polish learners tended to avoid using collocations whose usage they were not convinced of, because the Polish teaching of English focused on accuracy. On the other hand, German learners tried to use alternative ways: paraphrasing and using definition, because they were used to taking English classes which were aimed at fluency of English communication. The other finding was that Polish learners were more dependent on L1 for producing English, while German learners looked for more creative strategies. This finding seemed to be related to semantic and formal similarity between L1 and L2 and the number of synonyms acceptable as collocates.

The purpose of Bahns and Eldaw (1993) was to investigate 58 German advanced EFL students’ productive knowledge of English collocations including the ability to paraphrase them in a translation task and a cloze task. They had two main findings: (a) Collocations were a major problem in writing English for advanced students. (b) Students had particular difficulty paraphrasing them. Based on these findings, they suggested that EFL teaching should focus on those collocations which are difficult for learners to paraphrase.

As a significant sub-purpose of their collocation research, Bahns and Eldaw (1993) also examined the relationship between collocational competence and general vocabulary size. They could draw the tentative conclusion that “learners’ knowledge of collocations did not develop in

parallel with their knowledge of vocabulary” (1993, p. 109). This might be in part due to the fact that collocations were not highlighted, considering how much vocabulary teaching was focused on. This emphasizes the need for effective collocation teaching.

Then, further questions were discussed by Bahns (1993): “Which of the tens of thousands of collocations do we select for special treatment in the classroom?” “And are there any criteria to decide which collocations need to be taught and which do not?” (1993, p. 58). Using a contrastive approach to the concept of lexical collocation, he found that the collocations, for which there are no direct translational equivalence, need to be focused on in the teaching of English to speakers of German and as a result, the enormous teaching and learning load can be reduced. In other words, in order to develop students’ collocational knowledge based on this idea in their classes, a workbook which presents a selection of collocations geared to their specific difficulties with a particular L1 background should be provided, as Mackin (1978, p. 151) also suggests.

Nesselhauf (2003) also maintains that the collocations for which there are no direct translational equivalence should be focused on in her research. She reported an exploratory study that analysed the use of verb-noun collocations such as take a break or shake one’s head used by advanced German-speaking learners of English in free written production. About 500-word argumentative and non-technical writing by third or fourth year German-speaking university students of English was evaluated in terms of their acceptability. In her research, whether the combination written by the students is a free combination, a restricted combination or an idiom is determined by several native speakers of English, dictionaries such as the

Oxford Advanced Learner’s Dictionary (OALD, 2000) and the Collins COBUILD English Dictionary (CCED, 1995) and the British National Corpus (BNC). The application to the data resulted in a database of verb-noun combinations classified as to their degree of restriction: free combinations, restricted combinations, more restricted combinations, idioms and to their degree of acceptability: clearly acceptable, largely acceptable, largely unacceptable and clearly unacceptable.

One of her results is similar to ones of Bahns (1993) and Biskup (1992) that the L1 influence on the production of collocations is rather high. The other result, her original one, shows that among verb-noun combinations the highest rate of mistakes occurs in combinations with a medium degree of restriction because of the wrong choice of verb on verb-noun combination mistakes. Therefore, the suggestion was made that the focus should be on the verb in the teaching of verb-noun collocations, since it is the verb that causes the greatest difficulties. Oikawa (1993) similarly recommends that core meaning of basic verbs should be taught, especially, for lower-level learners because they face difficulty in producing verb-noun collocations because of their L1, Japanese. Her other suggestion is that collocations should be taught with reference to L1 and it is important to teach them including prepositions, articles, etc, because of the difference between L1 and L2.

Nesselhauf (2003) implies that three criteria for selection of collocations should be considered. One is acceptability and frequency (p. 238).

Collocations to be learned should be acceptable and frequent in a neutral register and any special register that is of use to the learner. The other two, which are more focused on by her, are restriction and congruence. As for

restriction, verb-noun combinations with a medium degree of restriction are the most frequently mistaken and so they should receive particular attention of learners. With respect to congruence, as non-congruence in L1 and L2 causes learners’ errors, non-congruent collocations should be more focused on in language teaching. In addition, she found in her data analysis that even congruent collocations were mistaken, which is opposed to Bahns’

argument (1993) that congruent collocations can be entirely disregarded because learners will automatically acquire them. Therefore, she strongly suggests that further research is needed on which congruent collocations are difficult for learners.

Caroli’s research (1998) also focused on lexical collocations in terms of L1 and L2 difference on the acquisition of collocations. Her aim is to examine the influence of L1 on English collocational knowledge and the learning burden for learners, and illustrate the relation between the development of learners’ knowledge of English collocations and their L1 vocabulary, and between the development of receptive and productive knowledge of collocation. Seventy-three Italian learners of English at upper secondary schools were asked to take Nation’s vocabulary test (1990), productive collocation test and receptive collocation test consisting of 15 Italian literal equivalent collocations and 15 non-equivalent collocations.

One of Caroli’s results, which is identified with that of Bahns and Eldaw (1993) but which is opposed to that of Gitsaki (1999), is that no close correlation was found between the general vocabulary knowledge and the collocational knowledge. In other words, even if general vocabulary increases steadily with the level of language learning, it does not necessarily mean that collocational knowledge increases at the same rate. Another result is

that at all stages of language learning, learners seem to have higher receptive knowledge of collocations than productive knowledge of collocations. The other result, which is related to L1 and L2 difference, is that learners at the early stage of vocabulary development resort to L1 features in selecting possible English collocates and gradually come to comprehend the correct target language feature. Furthermore, Italian literal equivalent collocations are more easily acquired, which provides less burden for learners than non-equivalent collocations, as many researchers mention.

As many researchers suggest as pedagogical implications, Caroli (1998) also proposes that a word should be presented with some of its most frequent collocates and that teachers should have the responsibility to create conditions which effectively compensate for the foreign learners’ lack of collocation learning.

Granger (1998) investigated the difference in productive collocations and formulae between native speakers of English and French learners of English based on the International Corpus of Learner English (ICLE). Above all, in the first part of her investigation, she focused on the collocational study of amplifiers functioning as modifiers of adjectives such as totally or highly, proposing the hypothesis that learners would use them less frequently than native speakers. She found that the frequencies of two amplifiers, completely and totally, which were direct translation equivalents in L1, were much higher than those of other amplifiers. This might be because by using amplifiers which are similar to the equivalent in French, they did not need to become risk takers and could avoid making mistakes. In short, L1 played an important role in collocation acquisition and use in the L2.

From the pedagogical implications of her research, Granger suggested

that EFL material should be developed based on this result to give learners the most efficient learning aid. She also proposed that EFL teaching programmes should not rely much on first language acquisition theory, citing Cowie (1998) that “the child first acquires ‘chunks’ of language, then analyzes them and finally develops from them regular syntactic rules” (p. 14).

This is because there is not enough evidence to apply the theory to the collocational acquisition for L2 learners. She concluded that empirical work was urgently needed to investigate the mechanism of collocation acquisition, using both the wide variety of large computer corpus and the introspective tests.

3.2.2. Collocation research in terms of several factors