Near-synonyms refer to words of similar or nearly the same meaning yet, which still have subtle differences. Many EFL learners have been confused by them and encountered difficulties in distinguishing usages (Yeh, Liou & Li, 2007; Zhuang, 2010).
This study aims at exploring semantic features of the selected near-synonyms group: by drawing a comparison of the collocational and colligational information revealed by the EFL learner corpora and the native corpus.
In order to identify semantic patterns of the selected verbs, only content word collocates were included in the study. This was to avoid the analysis being cluttered with very high frequency collocations such as Understand his, or Understand to. The main categories of function words excluded from the analysis included personal possessive pronouns, cardinal numbers, determiners, prepositions, and adverbial particles.
Following the guiding principles of analyzing near-synonyms, there are two steps for data analyses. First, all the reference dictionaries were consulted to explore whether there was any explanation on the usage and phraseology of the synonym sets. The second part of the analysis focused on a survey of the different collocates between them by COCA and Sketch Engine. Given that these near synonyms have similar meaning, one might have expected a significant similarity in the type of nouns in the subjects. A useful tool, Compare Words in COCA, allowed for a comparison of the collocates of two near-synonyms, such as Small/Little, or Start/Begin, which provided insight into the differences in meaning or use of these two words. This function would assist in analyzing the semantic pattern or the phraseology of the target near-synonym sets. For example, Figure 12 showed the search result of two near-synonyms: Finish and End.
Figure 12. A comparison of collocates with Finish and End by COCA.
The basic idea of the table on the right column of Figure 12 was a list of the collocates of two competing words (i.e., near-synonyms in the present study) in a descending order of frequency. The tool Compare in COCA allows for a comparison of semantic features of the synonym sets. A screenshot of the top 20 collocates of the two synonyms was presented in Figure 12. According to Mark Davies, the founder of COCA, the basic idea of the Compare function is that we want to see how frequent a collocate occurs with two competing words, compared to the overall frequency of those two words. For example, if there are twice as many tokens of Word1 as Word2 in the
corpus overall, but a given collocate occurs fifty times as much with Word1 as with Word2, then the ratio of Word1 to Word2 with that collocate is 25 times what would otherwise "be expected." In Figure 13, the collocate job, the ratio of [Finish/End] (25.4 times as frequency) is 80 times the overall ratio of [Finish/End] in the corpus. It is therefore found that job tends to be used with Finish than End, though the two verbs are near-synonyms.
Figure 13. A screenshot of the top 20 collocates of Finish/ End by COCA.
The number of collocates was set to 100, so we could review all the 100 collocates and pay attention to those with a particular high score and divide them into different semantic fields. This categorization would assist in differentiating near-synonyms which dictionaries are unable to achieve. In this example, more collocates of Finish/End were compared and listed in Figure 14. We discovered that dinner, term, homework, painting, coffee, and task went with Finish more frequently while lives, occupation, fighting, discrimination, marriages, suffering, crisis and wars occurred with End. This finding was consistent with what Kennedy (2008) had explored the difference between Finish and End. Drawing on a whole BNC data, Kennedy (2008) undertook two separate analyses for the verbs. The first was to count the frequency of a word as a collocate of the verb and only content words of collocates were included. The second part of analyses was to calculate Mutual Information values (MI). The MI score was used to compare the actual frequency of co-occurrence of two words with the predicted frequency of co-occurrence of the two words if each were randomly distributed in the corpus (Kennedy, p. 23). The result of the analysis was that Finish was associated with more mundane or small-scale activities or events. In our example, the collocate dinner, homework, painting and task could refer to the small-scale activities. On the other hand, End tended to be followed by big, unpleasant processes which have negative effect. Figure 14 showed that fighting, discrimination, suffering, crisis and war directly fell into the category as Kennedy had suggested in his study.
Figure 14. A screenshot of the collocates of Finish/End by COCA.
Co-occurrence or collocation has been used as a search principle for differentiating synonyms (Partington, 2004; Renouf, 2009). Another supplementary tool to explore collocation to differentiate near-synonyms is Sketch Engine15. Sketch Engine, developed by Adam Kilgarriff, is a corpus query tool, for learning about word sketch, including collocation and colligation. It includes in a very large corpus of English or Chinese (or indeed any language), and uses the information to give summaries of how a word behaves: what its collocates are, and what contexts it
15 The URL of Sketch Engine is: http://the.sketchengine.co.uk/login/
appears in. The program has been used by both Macmillan and OUP in dictionary and thesaurus production, and for a wide range of teaching and research applications.
Figure 15 demonstrates the search of Happen for near-synonyms by Sketch Engine.
Figure 15. A search of near-synonyms of Happen by Sketch Engine.
And the result of near-synonyms search is shown in Figure 16. A list of
near-synonyms of Happen are generated in a descending order of significance score.
For example, the top three near synonyms are Occur, Exist, and Appear. Occur was selected for further analysis in the result section.
Figure 16. A search result of near synonyms of Happen by Sketch Engine
Unlike COCA in which the researcher has to set collocation span for a node word, Sketch Engine offers a summary of collocates of a node word under different categories:
such as subject, object, modifer etc as shown in Figure 17.
Figure 17.A search result of Happen/Occur as near synonyms by Sketch Engine.
Based on Figure 17, the frequent modifiers of Occur are naturally, commonly, spontaneously, simultaneously and frequently. This common pattern doesn‘t mean that Happen can‘t go with naturally but represents that the relative frequency of naturally Occur is significantly over naturally Happen (182, 9). The corpus survey demonstrates what significant collocates native choose to go with Occur instead of Happen.
Above all, both COCA and Sketch Engine are useful tools to differentiate near synonyms. To search for native corpus data, COCA is larger in corpus size while Sketch Engine only has access to British National Corpus. However, both corpus query tools
assist in both collocational and colligational analyses to provide a detailed description of a node word. Sketch Engine also provides a Chinese interface and automatically divides the collocates into different categories: subject, object, modifier etc. The significant collocates for each verb are marked with different color. This helps corpus researchers to identify the frequent or specific collocates for a node word. In summary, to describe the differences among the near-synonyms in the present study, the information for collocation and colligation under the search result of COCA and Sketch Engine will be reported.
C
HAPTERF
OUR: R
ESULTS ANDD
ISCUSSIONThis chapter presents the results of three analyses: transitivity errors, contrastive interlanguage analysis (CIA) on semantic prosody, and analyses on near-synonyms.
Following the presentation of the results will be an examination of the issues emerging from the analyses. They will be illustrated with examples from the reference corpus and two learner corpora.
Transitivity Errors
The six verbs selected for analyzing transitivity errors are Arrive, Agree, Care, Dream, Listen and Reach. They were chosen not only for their high frequency of use in the EFL learner corpora but also for their problematic use indicated in previous studies (Chan, 2004, Chen, 2002, Gui & Yang, 2003, Huang, 2007). Both quantitative and qualitative analyses will be presented. The former focuses on the frequency of transitivity errors in the use of the selected verbs, while the latter presents detailed analyses on collocation and colligation for specific verbs. Explanation of EFL learners‘
transitivity errors will be presented at the end of this section.
Quantitative Analysis
Table 11 shows frequencies of the verbs from the EFL learner corpora and the percentage of the target construction of those verbs. As shown in Table 11, students depended on the construction of Arrive, Care, Listen and Reach for almost half of the constructions that can be categorized as ―Verb+(prep.)+object(noun/noun phrase) construction.‖ Of those verbs, Reach accounts for the highest occurrence, up to 80%.
Compared to the previous verbs, the target construction was less frequently used for Agree and Dream. Each consists of only 25% for Agree and 19% for Dream. EFL learners seem to have other choices in addition to the target construction, such as
that-clause or infinitive phrase.
Table 11 A Summary of Occurrences “V+(prep) +Object” in the EFL Learner Corpora Agree Arrive Care Dream Listen Reach
Table 12 shows the number of transitivity errors for each verb, obtained with the coding procedures presented in Chapter Three. Typical difficulties were found in the misuse of intransitive verbs as transitive (Agree with/on, Arrive at, Care about/for, Dream of, Listen to/for) and transitive used as intransitive (*Reach to the place). As revealed by Table 12, EFL learners made more errors when they misused intransitive verbs as transitive ones. The only one type of error, misuse of transitive as intransitive (Reach), has the fewest errors. This phenomenon probably reflects a lack of knowledge about verb transitivity in English.
Table 12 A Summary of Transitivity Errors
Verbs Examples of misuses in transitivity Number of errors
Agree *agree me, *agree this point 73
Arrive *arrive to the temple, *arrive the town 58 Care *care her children, *care kid development 100
Dream *dreamed my grandfather/the sea 63
Listen *listen the student/the song/English program 202 Reach *reach to her, *reach to the limit 32
Figure 18 shows the percentage of transitivity errors on the selected high-frequency verbs. A more detailed analysis of how the selected verbs are used by EFL learners points to some interesting tendencies. A first observation is that the learners appeared to commit a moderate high error rate, ranging from 23% to 37.2%
except for Reach with an error rate of only 9%. EFL learners seem to be familiar with the transitivity of Reach. A second observation which can be inferred is that Arrive seems to be the most problematic for students, followed by Listen, Dream, and Care.
Figure 18. Error rate of transitivity in the EFL learner corpora.
The results on transitivity probed so far show convincingly that there are substantial error rates (at least one fourth except Agree and Reach). In the following section of analyses, a further comparison of error rate between the two learner corpora will be made.
A comparison of error rate from TLEC and CLEC is shown in Figure 19. Students in Taiwan have higher error rates in Agree, Arrive and Care, while students in China make just slightly more in Dream, Listen and Reach. The quantitative results showed that using an intransitive verb followed by a noun or noun phrase serving as an object is the most common and problematic error type. Detailed descriptions about how EFL learners misuse the verbs will be discussed in the next section.
Agree Arrive Care Dream Listen Reach
Figure 19. A comparison of error rate between TLEC and CLEC.
Qualitative Analysis
In addition to identifying and counting the transitivity errors, a detailed analysis on collocation and colligation was presented through qualitative analysis. The major source of error is misuse of intransitive verb as transitive and vice versa for Agree, Arrive, Care and Reach.
AGREE
Agree has 1117 occurrences in TLEC while only 142 CLEC. The huge discrepancy resulted from the recurrent writing prompt in TLEC, such as ―Do you agree or disagree with the following facts: People should sometimes do things that they do not enjoy with the view (104), Agree this view (106), Agree the view (23).
Agree Arrive Care Dream Listen Reach
prompt. Similarly, other similar repeated writing instructions17 involving the word Agree were removed from the analyses.
Agree has been reported to be problematic and confusing for EFL learners because it comprises intransitive and transitive patterns yet with different meanings (Gui and Yang, 2003; Huang, 2007). Figure 20 shows a random selection of concordances of Agree. The errors could be divided into three types: lack of a preposition for intransitive uses, wrong preposition, and confusion between transitive and intransitive uses. First, the erroneous occurrences often lack a preposition before a noun or noun phrase, as shown in line 7 and line 32:
[10]*However, I do not [agree this point.]18 (TLEC)
[11]*We wouldn‘t like to [agree the others opinions.] (TLEC)
Another example of lack of prepositions before a personal pronoun was given in the following:
[12]*And they [agreed me] to do the part-time job. (CLEC)
17 Similar writing instructions include, ―Do you agree with the following statement?‖
18 Please refer to Chapter 3 for Lennon‘s (1991) guidelines for identifying learner errors.
Figure 20. A random selection of the concordances of Agree.
The second type of error is attributed to the wrong preposition after Agree.
[13]* Personally, I don't [agree on those people] described above wholly19. (TLEC) [14]* I partly [agree to this subject]. (TLEC)
The third type of error causing learners‘ confusion about Agree is its dual structures as intransitive or transitive. The transitive patterns of Agree include ―Agree + N (NP),‖
―Agree + (that) + clause,‖ and ―Agree + what + clause.‖ Their patterns will first be evaluated referring to dictionaries and the reference corpus COCA. It is interesting to
19 In order for smooth readiability, the spelling errors in students‘ concordances were corrected
find out that the dictionaries offer confusing or even contradictory examples in this pattern.
If Agree is used transitively and followed by a noun object, it means to ―decide together‖ as shown by the examples given in the Longman Dictionary of Contemporary English: Agree a price/plan/strategy. However, the same object ―plan‖ was found to be used as intransitive in online Oxford Advanced Learners’ Dictionary: ―Agree to a plan.”
Only Collins Cobuild Advanced Learner’s English Dictionary (CCALED) explained that if people Agree on something or in British English if they Agree something (used as a transitive verb), they will decide to accept to do something. One examples given from CCALED is ―We never agreed a date.‖ The above explanations demonstrated that the transitive pattern of Agree appeared more often in British English than in American English. The native corpus search showed that if Agree is used transitively, it is very often followed by a that-clause, or a wh-clause. The direct object after Agree, though written in the dictionary, is rarely found in the native corpora.
ARRIVE
As shown in Figure 19, of all the selected verbs in the present study, Arrive has the highest error rate of 37.2% and Taiwanese students committed 46% of the errors. The major error type was the omission of a preposition when Arrive was used intransitively.
Many errors were found in the structure: ―Arrive + at + (place).‖ Learners were not aware of the intransitive use of Arrive and omitted the preposition ―at‖ before an object.
Typical errors were shown with the following examples:
[15]*When he [arrived the classroom], he may say to the teacher. (TLEC)
[16]*Players go to their lengths to try to be the first one [arriving the terminal.]
(CLEC)
Even if learners used the correct intransitive structure, they might use the wrong
prepositions. Students seemed to be much confused by the intransitive structure. In TLEC, there were several erroneous occurrences such as *arrive to the temple/school;
*arrive at Hong Kong/California. More examples were given in the following:
[17]*The first top of our tip, we [arrived upon Vancouver] which is a beautiful city.
(TLEC)
[18]*But at half past ten, two men from the government [arrived to professor‘s house.](CLEC)
What might be more confusing for students was that a preposition could be unnecessary in the phrase ―Arrive + home.‖ However, at was required when there was possessive, such as ―Arrive + at + (someone‘s) home.‖ Some students mixed both and made the problematic patterns:
[19]*When I first [arrived at home], a special feeling came to my mind. (TLEC) [20]*On that day, most relatives will [arrive our home] and we enjoyed this special and rare time together. (TLEC)
In addition to lack of prepositions and confusion about prepositions, there were many collocation errors with Arrive from the EFL learner corpora. Some examples were given below:
*arrive at this object/goal/your purpose 達成目標/目的 [correct use: achieve the goal/purpose20]
*arrive any achievement 達到..成就
[correct use: increase/promote the achievement]
*arrive the real functions 達到..功效[correct use: improve/perform the function]
20 The correct collocation were checked by COCA or Tango. Tango, a collocation query tool with its URL at http://candle.fl.nthu.edu.tw/collocation/webform2.aspx?funcID=9/
CARE
Of all the 356 occurrences of intransitive use of Care, 100 errors were identified.
The errors of Care can be divided into three types. The first type is consistent with the previous verbs: misuse of intransitive as transitive with a lack of preposition (for, about) before the object. The following examples illustrate some typical errors of Care:
[21]*She really [cares her children], just like normal parent. (TLEC)
[22]*He can observe and [care every employee's life matter] and task trouble and come to help him. (CLEC)
The second type of error is confusion about prepositions. Most students are familiar with the phrase ―take care of‖ but misused the preposition when Care is used as a verb. Specific examples were given below:
[23]*They are more independent because they wanted to [care of their younger siblings]
as parts of parents. (TLEC)
[24]*My uncle was so [care of my condition.] (TLEC)
[25]*No body [care of him], the car ran around his body and seemed likely run over him. (CLEC)
In addition to the two error types, a further colligation analysis revealed that there were differences in the structures between the two learner corpora. Table 13 showed that the transitive pattern of ―Care+if/what/whether+clause” was more commonly used in TLEC (45 occurrences) than in CLEC. There was only one concordance of the structure in CLEC:
[26] They won't care whether I'm in good mood or not and I won't trouble myself to find out what they really think or feel. (CLEC)
Table 13 Colligational Analysis Between CLEC and TLEC
Category CLEC TLEC
(1) care about/for something/somebody 131 156
(2) I don't care 4 3
(3) to-clause 0 1
(4) if/what/whether/that clause 1 45
Total 136 205
One possible reason for the difference is probably that CLEC contains about 19%
of learner data from writings by high school students, who might not be familiar with the complex clause structure after Care. Another reason could be topic-related. There are more essays in TLEC that are related to personal views on different subjects, and students use Care to express their ideas. For example, in an essay about real friendship, the student wrote,
[27] Only friends don‘t care whether you are rich or not. They don‘t care if you take advantage of them. (TLEC)
REACH
Of the six selected verbs, Reach has the lowest error rate in transitivity (9%). The error type was consistent with the wrong use of the preposition ―to‖ as found with Reach from the learner corpora. There were some error patterns identified with Reach.
For example, if the object of Reach was a place or goal, most students used the transitive structure correctly. However, if the object was related to number or age, students made mistakes by adding the preposition ―to‖ after Reach as shown in examples 28 and 29:
[28]* The drop of price between the two kinds of monitors would [reach to two thousand.] (TLEC)
[29]*With the development of the society, that is with people's living conditions bettering, people more concerning about their health then taking part in all kinds of
[29]*With the development of the society, that is with people's living conditions bettering, people more concerning about their health then taking part in all kinds of