Available online at www.sciencedirect.com
System 38 (2010) 391e401
www.elsevier.com/locate/system
The relationship between vocabulary size and depth for ESP/EAP learners Is’haaq Akbarian Department of English Language and Literature, University of Qom, PO Box No. 37185-396, Qom, Iran Received 16 December 2009; revised 8 February 2010; accepted 1 June 2010
Abstract Vocabulary knowledge occupies an important position in language learning. This study investigates the relationship between vocabulary size and depth for Iranian learners of English for specific/academic purposes (ESP/EAP). The participants include 112 ESP graduate students at a university in Iran. The findings from linear regression analyses show that, overall, VLT (size test) and WAT (depth test) have a great deal of common shared variance for these participants (R2 ¼ .746). However, when they were divided into low and high proficiency groups, based on whether the participants mastered the most frequent 2000 words in VLT, a substantial amount of shared variance was shown for the low group (R2 ¼ .464) and a much higher one for the high group (R2 ¼ .804). The findings suggest that vocabulary size and depth might be accounted for by the same factors, especially as the learners’ proficiency increases. Ó 2010 Elsevier Ltd. All rights reserved. Keywords: Vocabulary size; Vocabulary depth; Relationship; ESP/EAP learners
1. Introduction Vocabulary is one of the most essential components of language learning. Accordingly, foreign language (FL) and second language (SL) learners are typically conscious of the extent to which limitations in their vocabulary knowledge affect their communication skills since lexical items carry the basic information they wish to comprehend and express (Nation, 2001). Researchers in the field of vocabulary learning and teaching have made a distinction between two dimensions of vocabulary knowledge: size and depth (Bogaards and Laufer, 2004; Haastrup and Henriksen, 2000; Milton, 2009; Read, 2000). However, most recently, reviewing a large number of studies in his excellent volume Measuring Second Language Vocabulary Acquisition, Milton (2009) empirically argues that these two dimensions are not separable and that they might be closely related. Size of vocabulary knowledge is considered as referring to the number of words that language learners know at a particular level of language proficiency (Nation, 2001). Researchers have used various types of assessment tools with different formats to measure this dimension of vocabulary knowledge (see Wesche and Paribakht, 1996, for a discussion of these various assessment types). Nassaji (2004) states that one widely used measure to assess the size of E-mail address:
[email protected] 0346-251X/$ - see front matter Ó 2010 Elsevier Ltd. All rights reserved. doi:10.1016/j.system.2010.06.013
392
I. Akbarian / System 38 (2010) 391e401
vocabulary knowledge in the literature is Vocabulary Levels Test (henceforth VLT), which has a wordemeaning matching format and is composed of words representing different word-frequency levels, ranging from highfrequency (2000-word level) to low-frequency words (10,000-word level). As Milton (2009) states, through these tests “we get believable and stable results” (p. 11) and they have good content validity (see Method section for an example item of VLT). Depth of vocabulary knowledge refers to how well the language learner knows a word (Read, 1993, 2000). According to Nassaji (2004, p. 112), researchers have indicated “the complexity and multi-dimensionality of word knowledge and have suggested that knowing a word well should mean more than knowing its individual meanings in particular contexts.” Various kinds of knowledge are associated with a word that a learner must know, ranging from knowledge of its pronunciation, spelling, register, and stylistic and morphological features (Haastrup and Henriksen, 2000; Nation, 1990; Richards, 1976) to knowledge of the word’s syntactic and semantic relationships with other words in the language, including collocational meanings and knowledge of antonymy, synonymy, and hyponymy (Chapelle, 1994; Henriksen, 1999; Read, 2000). One widely used measure assessing only some of these aspects is Word Associates Test (henceforth WAT) that was originally developed by Read (1993, 2000). He used the principle of word association to make a word associates format that asks learners to choose responses to a stimulus. The target word and associates have three basic relationships: “paradigmatic (superordinates, synonyms), syntagmatic (collocates) and analytic (words representing a key element of the meaning of the target word)” (Read, 2004, p. 221). The instrument was refined through a process of repeated piloting. The test has been found to be closely correlated with SL reading comprehension ability and has also been shown to have a high degree of internal reliability (Qian, 1999, 2002; Read, 1993). WAT measures only some components of vocabulary depth, as pointed out above, since these components are important. They appear frequently in discussions of vocabulary knowledge (e.g. Chapelle, 1994; Nation, 1990, 2001; Qian, 1999, 2002; Read, 1993, 2000; Wesche and Paribakht, 1996). Although the test taps knowledge of adjectives only, given the design of the measure that requires the identification of nouns that collocate with the adjectives tested, nouns are actually indirectly tested as well (see Method section for an example item of WAT). What is worth mentioning concerning depth is that there are “no clear, comprehensive and unambiguous definitions to work with” (Milton, 2009, p. 150). The construct validity of depth is therefore challenged. However, one approach is to test the elements of vocabulary depth separately, say idiom or collocation, for the assumption that it will represent ability in the whole of vocabulary depth. Given that, Wolter (2005) is not so optimistic about this test and cannot suggest with confidence that it succeeds in assessing depth of word knowledge. In addition, “[g]uessing and vocabulary size are likely to play a significant role in the scores the test produces (Milton, 2009, p. 163). 1.1. Challenges and contextual constraints initiating the study The Iranian context of English for specific or academic purposes (ESP/EAP) places the priority in language learning on reading skill, since ESP/EAP learners at tertiary level need to study large amounts of text in English for their studies. This approach leads to more emphasis on grammar and vocabulary components of language learning. However, in teaching vocabulary for ESP/EAP at tertiary level, no systematic approach is adopted to help the university students notice sufficiently the dimensions of vocabulary knowledge, as expected. This kind of emphasis is not unique to Iran, though. It seems to be common throughout the world (Chui, 2006; Grabe and Stoller, 1997; Nurweni and Read, 1999). Another factor adding to Iranian students’ poor proficiency in English is that, rather than highlighting the role of international language in the development of the nation as well as pushing and pulling ESP/EAP learners at the university to learn English systematically, the faculty members of Iranian universities resort to translating technical textbooks into Farsi. So ESP/EAP learners do not see the need to study an original textbook in English. This is rather disappointing since the university students get limited to and, as a consequence, dependent on Farsi textbooks and do not study English textbooks independently out of the curriculum. Of course, the researcher agrees with authoring technical textbooks in Farsi for any field, but does not endorse only translating English technical books into Farsi. Unfortunately, the latter, as Parviz (2008) observes, goes ahead in Iran. Under such circumstances, increasing the size of vocabulary knowledge of EFL or ESP/EAP learners through explicit teaching is difficult within the scope and the time in which a language learning program is offered. Accordingly, it is doubly difficult to explicitly teach or highlight aspects of the quality of the lexical items.
I. Akbarian / System 38 (2010) 391e401
393
Nonetheless, researchers (Nation, 2001; Read, 2000; Schmitt, 2008) suggest that not only do learners need a large number of lexical items, but they must know a great deal about each item as well so that they can use the words well. Since vocabulary learning is a great hurdle for ESP learners and the learners do not seem to learn even the required level of vocabulary size through and within the scope of the offered language programs, Schmitt (2008) suggests that “a more proactive, principled approach needs to be taken in promoting vocabulary learning, which will require contributions from four learning partners”. The four vocabulary learning partners include students, teachers, materials writers, and researchers. He believes that “students need the willingness to be active learners over a long period of time”. Teachers need to provide guidance about which lexical items to learn and develop effective learning techniques. The expertise and resources of the researcher are necessary “in providing reliable information about vocabulary itself (such as frequency lists), and effective methods of learning it”. Materials writers should deliver “this research-based information to teachers and learners in a form that is usable” (p. 333). Schmitt asserts that the failure of any partner will end in the failure of the whole enterprise of vocabulary learning. Now, to link the issue at stake to the line of research on vocabulary knowledge and to put it in the right context, some relevant studies on vocabulary knowledge are discussed below. 2. Literature review 2.1. The relationship between size and depth The concepts of size and depth are not polar opposites. In fact, as a small amount of evidence suggests, these dimensions are somewhat closely related (Read, 2004). The tendency among the researchers, however, has been to contrast size and depth as two distinct dimensions of vocabulary knowledge. Therefore, this section reviews some available researches that investigate the relationship between size and depth as two distinct dimensions. This section also shows that non-native speakers, especially ESP/EAP learners, do not develop these dimensions well and equally. Administering VLT and WAT to 44 Korean speakers and 33 Chinese speakers, Qian (1999) found that the scores of the two tests were closely and significantly correlated at .78 for the Korean speakers and .82 for the Chinese speakers. He concludes that size is as valuable as depth to vocabulary knowledge since these two dimensions overlap one another and are interconnected. Qian (2002) also observes that the “scores on the depth and size of vocabulary knowledge measures are both capable of explaining a considerable portion (over 50%) of the variance in reading comprehension scores” (p. 532). Adopting a network building perspective on depth, Vermeer (2001) explored the relation between size and depth of word knowledge and linked these concepts with language acquisition and frequency of language input. In the first study, the size and depth of word knowledge of 50 Dutch monolingual and bilingual kindergartners (average age ¼ 5.6) were studied using receptive vocabulary, description, and association tasks. In the second study, the researcher investigated the relation between the probability of knowing a word and the input frequency of that word in 1600 Dutch monolingual and bilingual 4- and 7-year-olds. The results showed that there was no conceptual distinction between size and depth of vocabulary, and that size and depth of vocabulary knowledge were affected by the same factors for both groups of subjects. The results also showed very high correlations between both groups with regard to the probability of knowing a word, which was strongly related to the input frequency in primary education. Vermeer explains that the greater the number of words known, the deeper the knowledge of the words. Thus, “a child who knows more words also tends to know more about each word” as a result (p. 231). Administering a breadth test (a Dutch version of the Peabody Picture Vocabulary Test) and a depth test (a word associates test) to Dutch primary school children, Schoonen and Verhallen (1998) compared vocabulary breadth and depth tests with performance on two cloze passages, regarded as measures of reading comprehension ability. The results showed the breadth test correlated strongly with the depth test and that each vocabulary test contributed around 5e10 per cent to the prediction of the cloze scores. However, the depth test accounted for some additional variance in the cloze scores beyond that by the breadth test. In another research, Nurweni and Read (1999) studied the English vocabulary knowledge of first-year students at an Indonesian university. Using a sample of 350 students, they administered a word translation test to assess vocabulary size and a word associates test to measure vocabulary depth (at the level of the General Service List and University Word List, taken together as 2800 words). Overall, the two tests correlated at .62. However, once the students were divided into three groups on the basis of their general achievement in English, the correlations varied
I. Akbarian / System 38 (2010) 391e401
394
extremely according to the proficiency level. High group (10% of the sample) obtained a correlation of .81, Middle group (42%) got a correlation of .43, and Low group (48%) produced a correlation of .18 between the two tests. More particularly, in a recent study, Chui (2006) conducted a study of Hong Kong university students’ English vocabulary knowledge, i.e. breadth and depth. With 186 participants, she employed the Productive Vocabulary Levels Test (Laufer and Nation, 1999) to assess vocabulary size and a self-constructed depth-of-knowledge test to assess lexical competence across different aspects. For the latter, Chui sampled 20 headwords from Coxhead’s (2000) AWL. For these words, the students 1) constructed a meaningful sentence, 2) identified part of speech, 3) gave an assigned derivative, 4) explained the meanings, and 5) selected one collocational word out of four options. The findings showed that the knowledge of high-frequency words was fairly high but that of low-frequency ones quite deficient. Also, the students recognized a reasonable range of academic words but the quality of the knowledge was unsatisfactory; in fact, they provided part of speech successfully, due to emphasis on teaching grammar in Hong Kong, but gave only one meaning to the words and not their extra meanings. They barely converted 52% of the headwords into the required derivatives and identified the correct collocational words for only 57% of the items. The detailed description of the studies above indicates that the two dimensions of size and depth might be quite closely related. In addition, the relationship between size and depth of vocabulary knowledge seems to increase with the proficiency level of language learners. Also, an interesting finding, emerging from the last study, is that the depth dimension might be lagging behind the size for non-native speakers. Given the concerns raised in the previous section concerning the Iranian context of ESP/EAP study and with regard to the findings of the studies cited in this section, it is necessary to know the size and depth of Iranian ESP/EAP learners’ vocabulary knowledge and whether they grow proportionately at the same rate, prior to taking any steps and making any effort to increase the vocabulary proficiency level of the learners at tertiary level. 2.2. Statement of the problem Although vocabulary has received increased attention in recent years, Milton et al. (2008) think that fresh vocabulary research can provide many contributions to the task of teaching and learning from the perspective of (a) understanding “how language is constructed, how it is learned, and how it is used in communication” (p. 135), (b) helping “to establish norms of progress and even standards of knowledge and performance” (p. 136), (c) “helping us to understand and control language input”, and (d) aiding teachers and learners to select “appropriate methodologies and techniques to enhance their progress and performance” (p. 137). Meanwhile, Vermeer (2001, p. 218) states that “too little is known about the relationship between these various aspects of word knowledge” (i.e. size and depth of word knowledge). Moreover, Milton (2009) calls for more research on vocabulary acquisition to produce more data and shed more light on the area so that a clear, comprehensive, and unambiguous definition of the concept of vocabulary knowledge is structured. The results of such studies can thus tell us about the accuracy and validity of theories of vocabulary learning that drive our teaching and testing. Such studies should include different learners, especially ESP/EAP students, who need to learn English for academic achievement and progress. Most of the studies done in Iran are mainly concerned with EFL learners, but not ESP/EAP students who need to read a lot in English in their fields. With all these in mind, this study intends to answer the following research question: Is there any relationship between vocabulary size and vocabulary depth for Iranian learners of English for specific/academic purposes? Based on the above question, the following null hypothesis is formulated: There is no relationship between vocabulary size and vocabulary depth for Iranian learners of English for specific/academic purposes. 3. Method 3.1. Participants To gather data, 112 Iranian graduate ESP/EAP learners, studying at the University of Qom, participated in this study. The participants included 48 males and 64 females who were majoring in physics, mathematics, and
I. Akbarian / System 38 (2010) 391e401
395
electronic commerce. These students were accepted at graduate level through a nationwide entrance examination in which all the undergraduates and undergraduate (senior level) students throughout the country take part. In general, the entrance examinations into undergraduate and graduate level for the different universities of the nation are based on a norm-referenced procedure. Given that the university under study is similar to many, if not all, of the Iranian universities, the English proficiency level of the graduate students could most probably be regarded as similar to that of the students at many other Iranian universities. Therefore, the results of this study might indicate an overview, though very small, of the English competence of the graduate students across many universities. Yet, we acknowledge at the outset that the number of our participants is limited and by no means represents the whole of the population. The performance of these participants is first reported as one group in the Results section. Then, their performance is considered with respect to their proficiency level. The participants are therefore divided into two groups of high and low proficiency based on whether they pass the cut-off score for mastering the most frequent 2000 words in VLT. This division puts 26 participants into the high proficient group and 86 ones into the low proficient group. 3.2. Instruments Two vocabulary measures were used in this study: Vocabulary Levels Test (VLT): VLT a) exists in terms of levels of frequency, b) consists of larger samples of words from different word-frequency levels, c) is statistically reliable (Read, 2000), d) is related to success in reading, writing, and general language proficiency as well as to academic achievement (Laufer, 1997) and can provide efficient placement and admission in language teaching programs. It appears to be practical, economical, easy to administer, and can be completed in a short time. Therefore, we used version 2 of VLT, revised and validated by Schmitt et al. (2001). Each level of the test contains 30 items. In a recent study with 306 EFL participating students, Akbarian (2008) reported a Cronbach alpha of 0.963 on VLT at the four 1000-, 3000-, 5000-, and 10,000-word frequency levels. The present study did not use the 10,000-word frequency level as being beyond the language proficiency of the Iranian ESP/EAP learners. The maximum possible score was 90, with one point for each item at the three levels (see Fig. 1 for an example of the items in VLT). Word Associates Test (WAT): Devised by Read (1993), WAT measures three vocabulary elements: synonymy, polysemy, and collocation. Most of the stimulus words are general academic adjectives. The reliability of the test (KR-20) is 0.92 (Read, 1993). The split-half reliability of the test in the study by Qian (2002) was 0.89. WAT contains 40 items. Each item in WAT consists of one stimulus word (an adjective), and two boxes, each containing four words. Among the four words in the left box, one to three words can be synonymous to one aspect of, or the whole meaning of, the stimulus word. Also, there can be one to three words that collocate with the stimulus word among the four words in the right box. The instruction sheet for the test taker further explains that there are always four correct answers in each Participants must choose the right word that goes with each meaning. They must write the number of that word next to its meaning. Here is an example. 1 business 2 clock 3 horse 4 pencil 5 shoe 6 wall
----- part of a house ----- animal with four legs ----- something used for writing
Participants answer it in the following way. 1 business 2 clock ---6--- part of a house 3 horse ---3--- animal with four legs 4 pencil ---4--- something used for writing 5 shoe 6 wall Fig. 1. A sample of vocabulary levels test.
I. Akbarian / System 38 (2010) 391e401
396
item. This arrangement effectively reduces the chances of guessing. In scoring, each word correctly chosen was awarded one point. The maximum possible score, therefore, was 160 for the 40 items. The following is an example: domestic home, national, regular, smooth
animal, movement, policy, speed
The scores obtained from this measure were treated as the variable of depth of vocabulary knowledge while those obtained from VLT were treated as the variable of size of vocabulary knowledge in the analyses. 3.3. Procedures and research design The participants were notified of the general purpose of the study and informed that their performance on the tests would not affect their course outcome. All the participants willingly took the tests in class periods. As for VLT, they were instructed to first read the six words and then the three definitions. The participants had to choose the right word that went with each meaning. The time allotted was 30 min. As to WAT, the participating graduate students were instructed to read each of the target words and then circle the four words closely related to the target word. The time allocated to WAT was 30 min, too. Linear regression analysis was performed to investigate the predictive power of vocabulary size (independent variable) on vocabulary depth (dependent variable) for all the participants as one group and for the participants divided into high proficient group and low proficient group based on whether they mastered the first most frequent 2000 words in VLT. Those mastering the first most frequent 2000-word level were considered as high group whereas those not mastering this level were regarded as low group. The cut-off score for mastering the most frequent 2000 words in VLT was taken as 24 out of 30 (N. Schmitt, personal communication, May 9, 2008). In fact, the participants passing the score answered 80 per cent of the items correctly. The overall alpha significance level was preset at p < .05 for all the statistical analyses, using SPSS 16.0. 4. Results As discussed above, to answer the research question, two considerations were put into practice: a) the overall performance of the participating students on VLT, and b) their performance on the test as high proficient group (mastering the first 2000-word level in VLT) and low proficient group (not mastering the first 2000-word level). For this reason, the results will be reported below in correspondence with this order. 4.1. Results for the participants as one group With respect to the overall performance of the participating students on VLT, the descriptive statistics in Table 1 provide a general profile of their achievements. It seems noteworthy that the scores were standardized, too. The means and standard deviations resulting from the standardized scores on the tests are also reported for easy comparison. Throughout this research report, in the tables for descriptive statistics, the means and standard deviations based on the standardized scores are abbreviated as SM and SSD, respectively, adjacent to the means and standard deviations based on (original) non-standardized scores. Prior to analyzing the predictive power of vocabulary size on vocabulary depth, the Pearson productemoment correlation coefficient of these two variables should be presented. Therefore, to compare the correlation between the tests, the Pearson correlation coefficient was conducted. The result shows a very strong positive correlation coefficient (R ¼ .864, p < .001) between VLT and WAT. Table 1 Descriptive statistics for the participants as one group.
WAT VLT
MPS
Mean
Std. deviation
SM
SSD
N
160 90
51.30 33.05
24.93 16.97
32.06 36.73
15.58 18.86
112 112
MPS ¼ Maximum possible score.
I. Akbarian / System 38 (2010) 391e401
397
Table 2 Model summary for VLT and WAT. Model
R
R square
.864a
1 a
.746
Adjusted R square
Std. error of the estimate
Change statistics R square change
F change
df1
df2
Sig. F change
.744
12.62243
.746
322.922
1
110
.000
Predictors: (Constant), VLT.
Simple or linear regression analysis was conducted to determine the predictive power of VLT on WAT. In the table (Table 2) of model summary for VLT and WAT, R ¼ .864, R2 ¼ .746. This suggests that VLT and WAT actually overlap one another to a large extent: VLT has about 75% explained variance in WAT and vice versa. To illustrate the percentage of increase in the independent variable and the resultant change in the dependent variable, Table 3 shows that we obtained a ¼ 9.380 for the intercept and b ¼ 1.268 for the slope. So, given the data, for each percentage of increase in WAT scores, the scores on VLT change b (1.268) units. The overall performance of the participants as one group produced very strong correlations for the scores on VLT and WAT. The overall result might, however, obscure the delicate relationship between these two dimensions for the learners with different levels of vocabulary proficiency. To show a clearer picture of whether the two dimensions correlate with one another across the levels, it is necessary to report the results of the high proficiency and low proficiency learners separately.
4.2. Results for the participants as high proficient group Table 4 provides a general profile of the participants’ performance in the high proficient group. Also, before analyzing the predictive power of vocabulary size on vocabulary depth, the Pearson productemoment correlation coefficient was performed to compare the correlation between the two tests. The result shows a very strong positive correlation coefficient (R ¼ .897, p < .001) between VLT and WAT for the high proficient group. To determine the predictive power of VLT on WAT, linear regression analysis was conducted. In the table (Table 5) of model summary for VLT and WAT, R ¼ .897, R2 ¼ .804. This suggests that there is a great deal of overlap between VLT and WAT: VLT has about 80 per cent explained variance in WAT and vice versa. The result for the high proficient group stands about 5 per cent higher in comparison with that shown for all the participants as one group.
Table 3 Coefficients for the participants as one group. Model
Unstandardized coefficients B
1
(Constant) VLT a b
a
9.380 1.268b
Standardized coefficients Std. error
Beta
2.620 .071
.864
t
Sig.
3.580 17.970
.001 .000
Dependent Variable: WAT. Predictor Variable: VLT.
Table 4 Descriptive statistics for high proficient group.
WAT VLT
MPS
Mean
Std. deviation
SM
SSD
N
160 90
83.62 55.23
14.64 10.68
52.26 61.37
9.15 11.87
26 26
MPS ¼ Maximum possible score.
I. Akbarian / System 38 (2010) 391e401
398 Table 5 Model summary for high proficient group. Model
R
R square
.897a
1 a
.804
Adjusted R square
Std. error of the estimate
Change Statistics R Square change
F change
df1
df2
Sig. F change
.796
6.60920
.804
98.664
1
24
.000
Predictors: (Constant), VLT.
Table 6 Coefficients for high proficient group. Model
1
(Constant) VLT a
Unstandardized coefficients
Standardized coefficients
B
Std. error
Beta
15.728a 1.229
6.956 .124
.897
t
Sig.
2.261 9.933
.033 .000
Dependent Variable: WAT.
To illustrate the percentage of increase in the independent variable and the resultant change in the dependent variable, Table 6 shows that we obtained a ¼ 15.728 for the intercept and b ¼ 1.229 for the slope. So, given the data, for each percentage of increase in WAT scores, the scores on VLT change b (1.229) units.
4.3. Results for the participants as low proficient group As for the way the participants in the low proficient group performed on VLT and WAT, Table 7 shows a profile of their achievements. The Pearson productemoment correlation coefficient was again conducted on the data to compare the correlation between these tests for low proficient group, showing a strong positive correlation coefficient (R ¼ .681, p < .001) between VLT and WAT. Linear regression analysis was conducted to determine the predictive power of VLT on WAT. In the table (Table 8) of model summary for VLT and WAT, R ¼ .681, R2 ¼ .464, suggesting that VLT and WAT actually overlap one another to a large extent. VLT and WAT have about 46 per cent shared variance. Accordingly, to illustrate the percentage of increase in the independent variable and the resultant change in the dependent variable, Table 9 shows that we obtained a ¼ 14.514 for the intercept and b ¼ 1.026 for the slope. So, given the data, for each percentage of increase in WAT scores, the scores on VLT change b (1.026) units. Comparing the Beta values under standardized coefficients (.864 in Table 3, .897 in Table 6, and .681 in Table 9), we find that VLT, as the predictor variable, makes a significant strong contribution to the prediction of WAT, as the Table 7 Descriptive statistics for low proficient group.
WAT VLT
MPS
Mean
Std. deviation
SM
SSD
N
160 90
41.53 26.35
18.24 12.11
26.96 29.28
11.40 13.46
86 86
MPS ¼ Maximum possible score.
Table 8 Model summary for low proficient group. Model
R
.681a
1 a
R square
.464
Predictors: (Constant), VLT.
Adjusted R square
Std. error of the estimate
Change statistics R square change
F change
df1
df2
Sig. F change
.458
13.42745
.464
72.772
1
84
.000
I. Akbarian / System 38 (2010) 391e401
399
Table 9 Coefficients for low proficient group. Model
1
Unstandardized coefficients
(Constant) VLT a
Standardized coefficients
B
Std. error
Beta
14.514a 1.026
3.483 .120
.681
t
Sig.
4.167 8.531
.000 .000
Dependent Variable: WAT.
dependent variable, for the three sets of analyses. However, the strongest contribution to explaining the dependent variable is for the high proficient group (.897). It means that the more the participants develop their size of vocabulary knowledge, the more their depth of vocabulary knowledge will increase. In the following section on discussion, we turn to interpret the results reported above. 5. Discussion In this article, the relationship between size and depth of vocabulary knowledge was investigated for 112 Iranian ESP/EAP learners as a combined group, and also divided into high and low proficient groups of learners with respect to whether they had mastered the first most frequent 2000 words in VLT. Before we attend to the correlation between size and depth of vocabulary knowledge driving the current investigation, it seems necessary to focus on the means and standard deviations first. As stated earlier in the Method section, given that the participants under study might represent many, but not all, of the Iranian graduate students, then the mean and standard deviation of the participants as one group (Table 1) might probably be indicative of the disappointingly low English competence of the Iranian graduate students in general and the participants under study in particular. The situation in some other EFL contexts seems to be similar to ours as our results are consistent with the findings of Nurweni and Read (1999) whose sample of study had achieved only 44 per cent of the General Service List and University Word List. The deficiency of word knowledge of our participants is not confined to the quantity or size alone, but the quality or depth of their word knowledge, likewise, suffers and even lags further behind their vocabulary size. Standardized means and standard deviations of the three sets of analyses clearly illustrate this phenomenon (cf. Tables 1, 4, and 7). The division of the participants into high and low proficient groups with regard to whether they had mastered the first most frequent 2000 words in VLT showed a clearer picture of the vocabulary knowledge of our participants; only 26 managed to pass the cut-off score for the 2000-word level in VLT. Out of these 26 participants in the high proficient group, only two of them passed the cut-off score for acquiring the 3000-word level in VLT. Thus, their level of vocabulary knowledge is not suitable at all to meet their needs. The situation is much worse for the participants in the low proficient group who did not manage to obtain the 2000-word level in the test (Table 7). In fact, the majority of the students in our sample fell in this group who have not passed the cut-off score for mastering the 2000 words of high frequency in VLT. In this respect, our results were similar to those of Chui (2006). It is noteworthy that our participants were graduate students whereas hers were undergraduate fresher students. However, the majority of the participants in Chui’s research were from Hong Kong who had attended English-medium schools while all our participants were like the minority of Chui’s study from mainland China with Chinese-medium education. In Iran, the medium of education is Farsi. Nonetheless, the participants in both studies did not obtain the 3000-word level. This finding is alarming since a threshold of the most frequent 3000 words provides at least 89 per cent of the lexical coverage of the texts (Nation, 2006). With 2000 words of high frequency, language learners or text readers might be familiar with only 86 per cent of the running words in the texts. According to Sutarsyah et al. (1994), a knowledge of 4000e5000 English words is needed to comprehend an undergraduate economics textbook. Some researchers (Laufer, 1992; Nation, 1990; Read, 2000) believe that even knowing 89 per cent of the words in a text does not guarantee the correct guessing of the meaning of the unfamiliar words in a passage. Read (2000) states that a familiarity with 95 per cent of the running words is the minimum for correct guessing, not to mention other studies (e.g. Hu and Nation, 2000) that suggest a higher threshold (98e99%) to read various types of texts. Therefore, the low vocabulary proficiency level of all of our ESP/EAP learners raises a great concern for their academic future and a formidable challenge for the language instructors.
400
I. Akbarian / System 38 (2010) 391e401
In addition, our findings corroborate Schoonen and Verhallen’s (1998) study in that the results show there is a very strong positive correlation between VLT and WAT for all the participants as one group and the participants in the high group, respectively producing 75 and almost 80 per cent shared variance for these two tests (Tables 2 and 5). It indicates that the process of vocabulary development with regard to the size and depth of vocabulary knowledge might be accounted for by the same factors for ESP/EAP learners to a large extent. As Vermeer (2001) concludes, there might not be a conceptual distinction between the two dimensions. The high correlation might be “a logical consequence of the fact that the lexical elements in the mental lexicon consist of interrelated nodes in a network, which specify the meaning of an element” (p. 231). Nevertheless, for all the participants as one group and those in the high group, there is respectively 25 and 20 per cent shared variance in WAT that is not accounted for by VLT or vice versa. Specifically speaking, to be on the safe side, the unexplained shared variance is somewhat considerable, implying that there are many other variables at work so that we cannot take the size and depth of vocabulary knowledge as overlapping one another completely in the mental lexicon. That is, the process of developing vocabulary size and vocabulary depth in Iranian ESP/EAP learners is not clearly and completely understood as a construct. The findings for the low proficient group indicate that there is a strong positive correlation between the two tests, with VLT having about 46 per cent shared variance in WAT or vice versa (Table 8). In comparison to the findings reported for the participants as one combined group (Table 2), the unaccounted variance for this group is about 54 per cent, that is, 19 per cent more than that of the participants as one group and 26 per cent more than that for the high group. This might imply that an ESP/EAP learner first learns a number of words, then, along with that he or she starts accumulating the network of word knowledge associated with them. It seems that learners acquire additional information about words once their basic vocabulary (i.e. size of vocabulary knowledge) has reached a certain point. This issue might be accounted for by the lower standardized means of WAT in comparison to those of VLT for the participants as one group and the participants in high and low proficient groups (compare Tables 2, 5, and 8). Thus, our study presents data in favor of Nurweni and Read (1999), suggesting in line with their argument that while breadth and depth of vocabulary knowledge might converge when language learners are relatively advanced, the dimensions are more distinct at lower levels of language proficiency, as Read (2004) comments. Though the low proficient group did not reach the threshold of 2000 words of high frequency, the correlation between the tests for this group (R ¼ .681) is still high. In spite of that, some researchers assume that it might be difficult to assess beginners’ or lower intermediates’ vocabulary knowledge with the use of word-frequency criteria (Richards et al., 2008). Yet, the high correlation obtained seems to show something about the growth of these dimensions. Thus, considering our findings so far, our research presents data in favor of rejecting the null hypothesis. In other words, we can conclude that there is a strong positive relationship between vocabulary size and depth for Iranian ESP/EAP learners. The implication of the finding might be that we do not need to teach size and depth of vocabulary knowledge separately. These two dimensions should be taught in combination in foreign language contexts. On the basis of the results obtained in this study, we might go further and state that we do not need to even test the two dimensions separately. A further interpretation of our findings might be that the correlation reported in the three sets of analyses points to the merits of both VLT and WAT. The findings suggest that the tests complement one another. In other words, they might provide complementary information about the current status of the vocabulary knowledge of ESP/EAP learners despite the fact that both tests have shortcomings. A different interpretation might be that WAT is not really a depth test. It is merely a breadth test masquerading as a depth test. For Milton (2009), association tests, such as WAT, are not successful in tapping the depth of vocabulary knowledge. Their major problem is that they are not checking the quality of associations and that breadth is likely to affect the scores they produce.
Acknowledgements This research was supported by the University of Qom (Registered No. 3245). I thank my colleagues, who kindly let me administer the tests in their classes, and all the participants involved in the study. Thanks are also due to John Read, University of Auckland, New Zealand, for providing the Word Associates Test. I am also grateful to the anonymous reviewer for providing very constructive comments on an earlier version of the paper.
I. Akbarian / System 38 (2010) 391e401
401
References Akbarian, I., 2008. The role of vocabulary knowledge in predicting performance on reading comprehension item types. Unpublished doctoral dissertation, University of Tehran. Bogaards, P., Laufer, B., 2004. Vocabulary in a Second Language. John Benjamins Publishing Company, Amsterdam. Chapelle, C., 1994. Are C-tests valid measures for L2 vocabulary research? Second Language Research 10, 157e187. Chui, A.S.Y., 2006. A study of the English vocabulary knowledge of university students in Hong Kong. Asian Journal of English Language Teaching 16, 1e23. Coxhead, A., 2000. A new academic word list. TESOL Quarterly 34, 213e238. Grabe, W., Stoller, F., 1997. Reading and vocabulary development in a second language: a case study. In: Coady, J., Huckin, T. (Eds.), Second Language Vocabulary Acquisition. Cambridge University Press, New York, pp. 98e122. Haastrup, K., Henriksen, B., 2000. Vocabulary acquisition: acquiring depth of knowledge through network building. International Journal of Applied Linguistics 10, 221e240. Henriksen, B., 1999. Three dimensions of vocabulary development. Studies in Second Language Acquisition 21, 303e317. Hu, M., Nation, P., 2000. Unknown vocabulary density and reading comprehension. Reading in a Foreign Language 13, 403e430. Laufer, B., 1992. How much lexis is necessary for reading comprehension? In: Arnaud, P., Benoit, H. (Eds.), Vocabulary and Applied Linguistics. Macmillan, London, pp. 126e132. Laufer, B., 1997. The lexical plight in second language reading: words you don’t know, words you think you know, and words you can’t guess. In: Coady, J., Huckin, T. (Eds.), Second Language Vocabulary Acquisition. Cambridge University Press, New York, pp. 20e34. Laufer, B., Nation, I.S.P., 1999. A vocabulary size test of controlled productive ability. Language Testing 16, 33e51. Milton, J., 2009. Measuring Second Language Vocabulary Acquisition. Multilingual Matters, Bristol, England. Milton, J., Daller, H., Malvern, D., Meara, P., Richards, B., Treffers-Daller, J. (Eds.), 2008. Vocabulary [Special issue]. Language Learning Journal 36, 131e134. Nassaji, H., 2004. The relationship between depth of vocabulary knowledge and L2 learners’ lexical inferencing strategy use and success. The Canadian Modern Language Review 61, 107e134. Nation, I.S.P., 1990. Teaching and Learning Vocabulary. Newbury House, New York. Nation, I.S.P., 2001. Learning Vocabulary in Another Language. Cambridge University Press, Cambridge. Nation, I.S.P., 2006. How large a vocabulary is needed for reading and listening? The Canadian Modern Language 63, 59e82. Nurweni, A., Read, J., 1999. The English vocabulary knowledge of Indonesian university students. English for Specific Purposes 18, 161e175. Parviz, M., 2008. Translation is a threat to publication market. (2008, August 30). JAM-E-JAM, 20. Qian, D., 1999. Assessing the roles of depth and breath of vocabulary knowledge in reading comprehension. The Canadian Modern Language Review 56, 283e307. Qian, D., 2002. Investigating the relationship between vocabulary knowledge and academic reading performance: an assessment perspective. Language Learning 52, 513e536. Read, J., 1993. The development of a new measure of L2 vocabulary knowledge. Language Testing 10, 355e371. Read, J., 2000. Assessing Vocabulary. Cambridge University Press, Cambridge, UK. Read, J., 2004. Plumbing the depths: how should the construct of vocabulary knowledge be defined. In: Bogaards, P., Laufer, B. (Eds.), Vocabulary in a Second Language. John Benjamins Publishing Company, Amsterdam, pp. 209e227. Richards, J.C., 1976. The role of vocabulary teaching. TESOL Quarterly 10, 77e89. Richards, B., Malvern, D., Graham, S., 2008. Word frequency and trends in the development of French vocabulary in lowereintermediate students during Year 12 in English schools. Language Learning Journal 36, 199e213. Schmitt, N., 2008. Review article: Instructed second language vocabulary learning. Language Teaching Research 12, 329e363. Schmitt, N., Schmitt, D., Clapham, C., 2001. Developing and exploring the behaviour of two new versions of the vocabulary levels test. Language Testing 18, 55e88. Schoonen, R., Verhallen, M., 1998, April. Aspects of vocabulary knowledge and reading performance. In: Paper Presented at the Annual Meeting of the American Educational Research Association, San Diego. Sutarsyah, C., Nation, P., Kennedy, G., 1994. How useful is EAP vocabulary for ESP? A corpus based study. RELC Journal 25 (2), 34e50. Vermeer, A., 2001. Breadth and depth of vocabulary in relation to L1/L2 acquisition and frequency of input. Applied Psycholinguistics 22, 217e234. Wesche, M., Paribakht, T.S., 1996. Assessing second language vocabulary knowledge: depth versus breadth. Canadian Modern Language Review 53, 13e40. Wolter, B., 2005. V_Links: a new approach to assessing depth of word knowledge. Unpublished doctoral dissertation, University of Wales Swansea.