This chapter presents the research method in six sections: the overall design, the participants, the instruments, the referencing tools, the experimental procedure, and the data analysis.
The Overall Design
In order to investigate the effects of using the general dictionary, the collocation dictionary and the collocation retrieval tool on students’ collocation learning, the present study chose: The Longman Dictionary of Contemporary English (LDCE, the online version), The Oxford Collocations Dictionary for Students of English (OCDE, the disc version), and the online collocation retrieval tool—Webcollocate (WEBC).
The participants involved were divided into three groups: one with OCDE, another with LDCE, and the other with WEBC. A translation test containing 30 collocation items was designed. To measure participants’ improvement in collocation production with the aid of the collocation referencing tools, a pretest, a treatment task, and a posttest were conducted. A questionnaire was also designed to investigate the participants’ perceptions of collocation learning and the use of referencing tools. The entire duration of this study lasted about four weeks.
63
Participants
One hundred and ten senior high school sophomores (three intact classes) participated in the present study. Their English proficiency level is around intermediate. They all have received at least eight years of English course in the formal education system. One class of students was assigned to use OCDE, another was assigned to use LDCE, and the other was assigned to use the online collocation retrieval tool—WEBC for collocation searching.
Instruments
Two instruments were designed and administered to collect data in this study: a translation test and a questionnaire. Detailed information is described as follows.
Translation test.
There were three tests conducted in the study, including a pretest, a treatment task, and a posttest. They were identical in the form but different in the item arrangement. The test was designed as a translation written test consisting of 30 V-N collocation items (see Appendix A). The chosen V-N collocations were found to be highly error-prone in Taiwanese EFL students’ collocation production based on the previous research (Chan & Liou, 2005; Lin, 2008). Each collocation was embedded in an English sentence with Chinese translation. The noun headword was given while the verb collocate was left blank. In addition, the headword was boldfaced to elicit the
64
participants’ knowledge of a proper verb collocate. Here is an example of the translation question:
The boy will enter college to ____ knowledge for his future career. (獲得知識) Questionnaire.
A questionnaire in Chinese was designed to survey students’ perceptions of collocation learning and the use of referencing tools (see Appendix 2). The questionnaire contained 11 items to probe into learners’ collocation learning behaviors, their perceptions toward using electronic dictionaries and retrieval tools for collocation learning. Except for one open-ended question about the strengths and the weaknesses of the assigned reference tools, ten items were designed with a 4-point Likert scale, ranging from 4 point, “strongly agree,” to “strongly disagree,” 1 point.
Here is an example of an item from the questionnaire:
我覺得使用牛津英語搭配詞辭典來學習搭配詞有助於我運用字彙的能力。
□ 非常同意 □ 同意 □ 不同意 □ 非常不同意
Tools
Oxford Collocations Dictionary for Students of English (OCDE).
Oxford Collocations Dictionary for Students of English, 2nd edition (OCDE;
2012) is a collocation dictionary based on the Oxford English Corpus, a database of about two million words of text in English taken from updated sources from around
65
the world. According to the introduction in its preface, OCDE contains more than 250,000 collocations for 9,000 headwords, with 75,000 example sentences. The wordlist of OCDE consists of nouns, adjectives and verbs, which are supposed to be the most commonly used. These 9,000 headwords are upper-intermediate students supposed to have already known, or expected to encounter as they move to a more advanced level of English. Most of the examples were based on the authentic texts in the corpus, with minor modification to make them more accessible to the users.
OCDE covers the following types of word combination:
Noun entries: adjective +noun, quantifier + noun, verb + noun, noun + verb, noun +
noun, preposition + noun, noun + preposition;
Verb entries: adverb + verb, verb+ verb, and verb + preposition;
Adjective entries: verb + adjective, adverb + adjective, and adjective + preposition.
The collocations of each headword are basically divided according to types of word combination; within each type, synonymous or semantically related collocates are grouped together ‘intuitively’—as the introduction claimed—and separated from each other by vertical bars. As a collocation dictionary, the OCDE does not give definitions to the entries, as normal dictionaries do, unless the entry has more than one meaning. Rather, it gives definitions of headwords only when regarded as necessary to distinguish different senses of the same word. For example, out of the 11 entries on
66
the first page—abandon, abashed, abbreviation, abhorrent, ability, ablaze, able, abolish, abortion, abscess, and absence, only 4 were given brief definitions: abandon,
ability, able and absence, as each of these four entries had more than one meaning.
Besides, only a very small number of collocates of the entries were given short explanations, and most of the collocations were not accompanied by example sentences. Even if there was any example following, the headword was always replaced with the “~” mark instead of appearing with its collocate as a complete collocation.
This edition of OCDE is accompanied by a CD-ROM, which claimed to offer pop-up definition and 30,000 word index to make searching easy. Compared with the print version, the disc version provides the same contents but offers a much clearer layout. First, the types of word combination were presented in red capital letters for users to scan quickly for the type they wanted. Second, each group of semantically related collocates was listed separately from each other with a big dot in front to mark the beginning of each group. Third, it is much easier to find the example sentences—if there is any—since each example sentence stretches independently in one line. However, the pop-up definition the CD-ROM claims to offer is exactly the same as that in the print version—no further definition or information about the headwords or collocations is given.
67
Take the searched results of the noun attention in the disc version of OCDE for example (Figure 1).
Figure 1. The search results of attention in OCDE
Two senses of the headword were revealed—the first: act of watching/ listening/
showing interest; the second: care. Each of the two senses covers different types of
collocations, such as ADJECTIVE, VERB + ATTENTION, ATTENTION + VERB, ATTENTION + NOUN, and PREPOSITION. In the first sense of the word, under the type of VERB+ATTENTION, 45 verb collocates are divided into 13 subsenses. In the
68
second sense, 8 verbs are divided into two groups and listed under the type of VERB + ATTENTION.
Longman Dictionary of Contemporary English (LDCE).
As a dictionary for general learners, Longman Dictionary of Contemporary Dictionary (the 4th edition, 2003) is based on Longman Corpus Network, a database of 300 million words from written and spoken British and American English from many sources, such as books, newspapers, conversations and advertisements. It contains 106,000 words and phrases, with 220,000 word combinations.
To help users remember the words they have looked up and understand the word within a context, this edition mainly focuses on two areas: examples and collocations.
The edition offers 155,000 natural examples for the 106,000 words and phrases—on average, at least one example for each word or phrase. These examples are usually slightly edited versions of real sentences from the corpus. Collocations are presented in bold type before or in the examples so that they are easy to find. If a word has lots of collocations, they are listed in the collocation box in color in the beginning of the entry, with examples following the box.
In this dictionary, all the headwords are defined by using the most common 2000 words in real life. Meanings in the entries are basically ordered in their
frequency in the language; that is, the most frequent meanings are placed first. Long
69
entries have signposts, highlighted in blue, to guide users to scan down the content of the entry and find the meaning they want. In addition, synonyms, antonyms, and related words are presented after the definition to help explain the meaning of the entry. Frequently-used collocations or word clusters are accompanied with
paraphrases to make it even easier for learners to understand.
An online updated version of LDCE provides 207,000 words and 220,000 word combinations for language learners to use. The layout of the online version is clearer than that of the print version. First, the number of each meaning unit is presented in red which is bright enough for learners to notice easily. Second, all the example sentences appear separately from one another, making them easier to read. To search for a certain word, just key it in the search box.
The results of the collocations of attention searched in LDCE online were presented below (Figure 2).
70
Figure 2. The search results of attention in LDCE.
71
Under the word attention, several different definitions were shown on screen (i.e.
1. Listen / look/ think carefully and 2. Interest). Each definition consists of the most common 2000 words. As the first meaning of the word has many collocations, they are listed in a collocation box, followed by examples for all the collocations in the box.
Each collocation in the example is presented in bold face to make it easier to identify.
After some collocations, such as turn your attention to somebody/something and give (your) attention to somebody/something, there are further explanations in parentheses.
Unlike the OCDE, which divides collocations into different types and present them according to their part of speech, the types of collocations presented by LDCE are mixed up in each sense. For example, in the collocation box, in the first sense of the entry attention, 4 types of collocations are shown: ATTENTION + PREP. (~ on sb/
sth, ~ to sth), V. + ATTENTION (pay ~, turn/ give your ~, keep sb’s ~), ADJ. +
ATTENTION (close/ careful ~), and ATTENTION + V. (~ wanders). Instead of scanning for the types of collocations first, users have to scan for the sense or definition they want and then look for the collocations needed.
WebCollocate--wikipedia edition (WEBC).
Webcollocate (WEBC) is a collocation retrieval tool based on Wikipedia downloaded from the web site. In WEBC, the related words or collocates of a certain word are listed according to their frequency in use instead of being classified or
72
clustered according to their meanings as what is presented in the above dictionaries.
Besides, neither definitions of the key words nor explanations of their use in context are given.
To search for the collocates of a certain word, on the front page of WEBC, the user first types in the keyword, specifying its part of speech and the sequence of the related word—whether it appears before or after the keyword. It is also necessary for the user to specify the part of speech of collocate that he is looking for, and the distance between the keyword and its collocate. Once the keyword is submitted, a list of all the possible collocates appears according to their frequency use in the corpus.
Next, the user clicks the collocate to explore more in detail.
Take the noun attention for example. When the user searches for a verb that appears before it, a long list of verbs will be listed on the screen. The top 25 collocates shown on the first page of the screen are: attract, draw, receive, pay, have, turn, gain, come, bring, catch, get, be, focus, give, call, garner, receive, need, devote,
divert, capture, direct, seek, do and distract—with frequency ranging from 2178 to 78.
Then, if the user clicks to view the content of the collocate attract, he sees hundreds of complete example sentences containing attract and attention, with the keyword attention highlighted in color red, making it easier to find. The wealth of example
sentences might intimidate the users, but, on the other hand, exposes them to authentic
73
materials. The rich information enables them to understand, or at least guess the meanings from the contexts.
Figure 3 is the first 13 example sentences containing the verb collocate attract that appears before attention in WEBC.
Figure 3.The search results of attention in WEBC.
74
Experimental Procedure
The overall procedure of the study included one pretest, one treatment task, one posttest, and a questionnaire. The whole procedure for data collection lasted four weeks. In the first week, the pretest was conducted. The participants had 30 minutes to answer the questions, without any assistance of the referencing tools.
In the following week, the treatment task was conducted. Before the treatment task started, the participants received a 15-minute guidance on the upcoming collocation task. After the instructions, the researcher provided step by step instructions to demonstrate how the participants could find the answer to the first example question: 1. It is worthwhile to _____ good habits. 培養習慣 by utilizing the referencing tools.
The demonstration was followed by two tryout examples (i.e. 2. A teenager was seen to _____ the car on fire and run away. 縱火 and 3. According to the mother, her
son could ____ no wrong. 犯錯) as hands-on experiences for the participants to make
sure that they understand how to use the assigned referencing tools.
75
After the instructions and demonstration, the participants went on with the treatment task for 45 minutes to search for the appropriate answers to the questions.
The participants were informed that, while they were carrying out this task, no other referencing tool was allowed except the assigned one. The answer sheets were then collected 45 minutes later and the participants were asked to fill out the questionnaire, offering their perceptions of collocation learning and the assigned collocation referencing tools. At the end of the experiment, the questionnaires were collected.
The posttest was conducted in the fourth week. The participants were given 30 minutes to answer the questions. Then, the posttest papers were collected and the correct answers to the test were given to the students in order to provide them with some positive feedbacks of this test.
Table 3 Data Collecting Procedure
Week Task Time
Week 1 The participants took the pretest without the use of collocation referencing tools to measure their collocation knowledge before the treatment.
30
Week 2 1. Brief introduction of the assigned referencing tool by giving three examples to demonstrate how the participants can use the tools to find the answers by using the tools.
2. Practice using the tool with two examples.
3. Fulfill the treatment task by using the assigned referencing tool.
76
4. Fill out the questionnaire about collocation learning and the use of the assigned referencing tool.
Week 4 The participants took the posttest without using collocation referencing tools
30
Data Collocation and Analysis
The current study collected mainly quantitative data and some qualitative data.
The primary quantitative data was the scores of one treatment task, two tests and the questionnaire. The total score of each test was 30 points. Each item response was scored on a scale of 0 to 1 point. Correct answer, with correct spelling particularly, was given 1 point; otherwise, 0 point was given. For the questionnaire, since ten items were designed with a 4-point Likert scale, each item was rated from 4 point, “strongly agree,” to “strongly disagree,” 1 point. The total scores for each participant’ responses were then calculated. The only qualitative data was the participants’ perceptions on the strengths and the weaknesses of the assigned reference tool from the open-ended question listed in the end of the questionnaire.
After the data was collected, two parts of analysis were conducted. One was the quantitative analysis. First of all, to examine the effects of using referencing tools on participants’ collocation production, the researcher used the repeated measure one-way ANOVA to analyze the three test scores done by each group. Second, with the independent samples one-way ANOVA, the researcher compared the scores of
77
treatment task to check which referencing tool was the most helpful in collocation searching. Third, to evaluate which reference tool enhances the retention of collocation better, the scores of the posttest were compared by adopting the independent samples one-way ANOVA. The researcher also applied the post-hoc analysis to examine the differences among the three groups. Last but not least, data collected from questionnaire was calculated and compared by using the independent samples one-way ANOVA. It was expected that the participants’ perceptions toward collocation learning and referencing tools could be displayed.
Besides quantitative data, qualitative data also needed to be analyzed. To further examine the participants’ perceptions of the application of the assigned referencing tools, the participants’ responses collected from the open-ended question in the questionnaire were analyzed.
78