The Effect of Word Class
on Lexical Development in L2:
The Case of Japanese EFL Adult Learners
VASILJEVIC, Zorana
第２言語習得において品詞が語彙習得にもたらす効果
―英語を外国語として学ぶ成人日本語話者の事例―
鷲麗
ワ シ レ美知
ビ ッ チゾラナ
Abstract: The study compared the effect of systematic, explicit vocabulary instruction on the acquisition of word meaning and collocations for nouns, verbs and adjectives. Data were collected from 18 high-intermediate level Japanese students of English. The findings suggest that word class may be one of the factors in the acquisition and availability of lexical items for productive purposes. In both receptive and productive tests of word knowledge, nouns were found to have a clear advantage over the other two classes. Adjectives were acquired more easily than verbs. Some collocation sequences were found to be under-represented in learners’ lexicons. Possible explanations for these patterns are reviewed and the pedagogical implications of the findings are discussed.
Key words : vocabulary teaching and learning, collocations, word class
INTRODUCTION
Research from L1 suggests that one important factor in the organization of lexical knowledge in the brain is grammatical category information. Studies of young children and aphasic patients suggest that phonological and orthographic
representations of different word classes are processed by independent neural mechanisms (Damasio, 1990; Hillis & Caramazza, 1995). The analysis of young children’s speech also revealed a bias towards certain word classes, which may be a result of specific properties of the input language. For example, in a verb-final language such as Korean or Japanese, children were found to use verbs and verb morphology earlier than nouns (Clancy, 1986; Choi, 1997). One possible explanation is that the position of verbs at the end of the sentence makes them more salient. Furthermore, in verb-final languages, the omission of nominal referents is common and single inflected verb utterances occur frequently, giving verbs greater phonological prominence. In English, however, nouns tend to be acquired earlier than verbs, accessed more easily, and be more resistant to loss (Goldfield, 2000; De Bleser & Kauschke, 2003).
English preference for nouns can be attributed to several factors. Many nouns denote concrete objects and are encoded as independent entities in the mental lexicon. They also have hierarchically organized semantic structures. Verbs, on the other hand, have lower imageability and are dependent categories connected to noun arguments. They often refer to temporary events and have multiple organizing principles, such as change, causality and manner (Miller, 1991). English is a systematic verb-second language and verbs typically occur in the middle of longer sentences. They are less likely to appear as single-word utterances than nouns, which makes them less salient (Goldfield, 1993; Tardif et al., 1997). Furthermore, verbs have more inflections than nouns. They are marked for tense, mood, aspect and for person and number agreement with a subject. As Goldfield (2000) points out, many of the frequently used English verbs tend to be irregular, with numerous variations in their form. Finally, an analysis of mother-child interaction indicates that maternal speech tends to elicit noun production but rarely prompts children to produce verbs. Children are often asked
to name things but rarely asked to name actions. Verbs in parental speech acts are used primarily to negotiate and manage a child’s behaviour by directing the child to initiate, maintain or discontinue an action. As a result, children tend to acquire nouns faster than verbs and have a more varied noun lexicon (Goldfield, 2000).
Adjectives lie somewhere between nouns and verbs. They can share semantic properties with verbs with the emphasis on transitory properties (e.g. obedient, daring) or, like nouns, denote more permanent, stable properties (e.g. huge, difficult) (Polinsky, 2004). Adjectives are interpreted either relative to noun arguments to which they are assigned, or together with verbs under the joint category of predicate. Therefore, in developmental terms, the noun/verb distinction is believed to have priority over the acquisition of adjectives as a distinct grammatical class (Berman, 1988).
While the acquisition of different word classes has received a lot of attention in the literature in terms of cognitive and linguistic development in L1, there are not many empirical studies which have examined the developmental sequences of different grammatical categories in L2. Pigada and Schmitt (2006) examined the acquisition of 70 nouns and 63 verbs from extensive reading of four graded readers in French. The experiment was conducted as a case study and it involved one subject. The target words were classified into six frequency groups according to the number of encounters in the texts. Three types of word knowledge were assessed: meaning, spelling and grammatical behaviour. The results of the study provided evidence of substantial word learning. Improvement, however, was not uniform across all three types of word knowledge. Spelling was enhanced in all noun frequency groups and four out of six verb groups. For meaning, limited improvement was observed for low-frequency nouns and verbs, with gains being slightly higher for nouns. Improvements in grammatical knowledge were recorded in all frequency groups of nouns. The gains for verbs, however, were much lower
than for nouns.
Kweon and Kim (2008) tested the effect of input frequency and word class on incidental vocabulary acquisition in the extensive reading of authentic texts. Twelve Korean learners of English were tested on their knowledge of vocabulary before reading, immediately after reading, and one month after the reading treatment. Nouns were found to be easier to retain than verbs and adjectives. Across all three classes, more frequent words were learned more easily. However, words of lower frequency were learned more effectively than words of a higher frequency when their meaning was crucial for text comprehension.
While the studies above provide some important insights into the nature of vocabulary learning in L2, they also have some limitations. The earlier studies focused primarily on the acquisition of word meaning and orthographic form. In Pigada and Schmitt’s study (2006), grammatical knowledge was limited to the use of articles with nouns and prepositions with verbs. However, word knowledge encompasses many other components such as associations, collocations, frequency, register, etc. (Nation, 2001). There are also some methodological concerns. For example, Kweon and Kim (2008) relied solely on self-reporting as a measure of students’ understanding of the target words on the pre-test and the first post-test. In addition, although their post-test data suggested that adjectives are acquired faster than verbs, the authors do not discuss the possible reasons behind this phenomenon, focusing their attention primarily on the differences between nouns and verbs.
The purpose of this study is to systematically examine the extent to which word class may affect the acquisition of meaning and collocational patterns in L2 beyond a 2000-word high-frequency level. Collocations describe how specific lexical items co-occur with other words in the lexicon. They were selected as a
criterion for vocabulary assessment as they are considered to be a pre-requisite for correct vocabulary use and communicative effectiveness (Lewis, 2000).
PREDICTIONS
The study focused on the acquisition of the three major lexical categories: nouns, verbs and adjectives. It was predicted that the three word classes would follow divergent developmental sequences due to conceptual differences and the different representations they have in the lexicon. The target nouns and their collocations were expected to be acquired faster than verbs because nouns tend to be more structured and more predictable. Nouns constitute constants with which verbs and adjectives are associated as functions (Berman, 1988). Verbs were expected to be learned more easily than adjectives. Adjectives are a kind of a ‘rhetorical device’ (Berman & Slobin, 1994) and are not as important as nouns and verbs for language comprehension or production. Therefore, it was predicted that they would be acquired later than those nouns and verbs with which they are associated.
PARTICIPANTS
The study involved 18 Japanese students of English enrolled at a private language school in Tokyo. The learners were at a high-intermediate level, with TOEIC scores between 770 and 830. They attended EFL classes once a week for a total of 3.5 hours (105 min of listening and 105 min of reading) for a period of 18 weeks. The study was conducted in the reading class.
The reasons why the study focused on higher-level learners are due to the specific quantitative and qualitative features of advanced language proficiency. In order to measure the potential differences in collocation patterns, it was important that learners should already have had a fairly large vocabulary size. A small
vocabulary size could potentially blur the distinctions in the number and nature of collocations produced for different types of headwords. Furthermore, the focus on high-level learners meant that the study could be integrated easily in the routine curriculum. In recent theory on L2 vocabulary acquisition, there is a lot of support for explicit instruction of collocations at advanced levels of proficiency (e.g. Lewis, 2000; McCarthy, 2007). This means that the vocabulary treatment adopted in this study was likely to meet the needs of the students at this level.
MATERIALS
Six articles were selected from English-medium newspapers and magazines such as Time, Newsweek and The Economist. An effort was made to ensure that a variety of topics were represented, so different student interests would be accommodated. Articles that required specialized knowledge were avoided. The selected texts were approximately between 1,400 and 1,800 words in length. Each article was studied over a two-week period.
TARGET WORDS
As the subjects were advanced students, and the study was conducted in a natural setting as a part of a regular coursework, it was important to select words which would be new to this group of learners.
The selection of the target words was done in three stages. First, one hundred and twenty words (twenty words from each article) that were anticipated to be new to the students were selected by the researcher. As each text was studied over two weeks, the target words were divided into twelve sets of 10 words each. An effort was made to ensure that all the selected words denoted a concept familiar to the learners, and thus the learners were expected to be familiar with the equivalents in their first language. This was an important consideration, as both
the pre-test and the post -test included a translation task.
While the above described criteria were appropriate for the selection of items that would become the focus of the in-class instruction, not all of the 120 words were considered suitable for the purpose of this study. Earlier studies showed that general (corpus) frequency and input frequency (the number of occurrences in the text) have an effect on the rate of lexical acquisition (Meara & Fitzpatrick, 2000; Pigada & Schmitt, 2006; Kweon & Kim, 2008). Zahar et al. (2001) identified input frequency as one of the factors conducive to learning, while Horst et al. (1998) found that text frequency played a much bigger role in word acquisition than corpus occurrence. Therefore, in order to objectively assess the effect that word class might have on lexical development, it was important to ensure that all target words belonged to the same corpus frequency range and had the same number of occurrences in the text.
In order to determine the general frequency of the selected words, Vocabulary
Profiler was used. This is online software which analyses a sample of text and
classifies the words into four frequency bands: the first thousand words (K1), the second thousand words (K2), an academic word list (AWL) and off-list words; that is all the words that do not belong to any of the first three categories. Out of 120 selected words 8 were classified as K2 words, 15 as AWL items and 97 words belonged to the off-list band.
The off-list band items were then examined for input-frequency. In order to ensure that text frequency did not bias the results, items with two or more occurrences were eliminated from the study. The final list included 78 items: 37 nouns, 18 verbs and 23 adjectives. A complete list of target items is provided in Appendix I.
PROCEDURES
The study had three parts:
a) A vocabulary pre-test which consisted of an English to Japanese translation of the target items and a collocation test.
b) Vocabulary treatment which included two homework and two classroom vocabulary activities.
c) A vocabulary post-test (English to Japanese translation of the target items and a collocation test).
1. Pre-test
Format. The pre-test assessed both learners’ receptive and productive
knowledge of the target items. Receptive knowledge was tested by asking the learners to translate English words into Japanese. The words were presented without context to prevent possible guessing. Translation into the L1 is an effective measure of receptive vocabulary knowledge, as it involves form recognition and precise meaning recall (Haastrup & Henriksen, 2000). This test format is also quick and simple to administer, enabling a large number of words to be tested in a relatively short period of time. Furthermore, translation into the L1 enables learners to express their understanding of the target word without being hampered by their lack of language proficiency in L2 (Read, 2000).
Productive knowledge was tested through a collocation test. The ability to use a word requires extended knowledge beyond word meaning. In order to produce language that is natural and accurate, learners must know which word combinations are acceptable in the target language. Meara (1996) proposed that beyond the 5,000-word level, vocabulary size becomes less important than the way in which vocabulary is structured in the learner’s mental lexicon. Similar conclusions were reached by Lewis (2000) and McCarthy (2007), who argue that
the main feature of advanced proficiency may be the ability to retrieve a repertoire of fixed expressions, rather than a large vocabulary size or grammatical knowledge. A collocation test was therefore considered a suitable measure of lexical proficiency for the target group of students. The general hypothesis was that learners with more developed vocabulary knowledge would have a more complex and more structured network of associations. The students were asked to write three English words that can go together with the target words. A sample of the portion of the test is available in Appendix II. The students were given 45 minutes to complete the test.
Scoring. On the translation test, each correctly translated item was awarded 1
point. To ensure reliability, students’ responses were examined by the researcher and one bilingual native speaker of Japanese. Where there was a disagreement, the responses were re-examined and scored by agreement.
On the collocation test, one point was given for each probable word combination. The test was examined by the researcher and another teacher, a native speaker of English. When there were differences in the ratings, the responses were re-examined. As a distinction between acceptable and unacceptable collocations is not always clear, and even native speakers may disagree on the typicality of particular word combinations (Schmitt, 1998), problematic responses were compared with norming information available in reference books such as Oxford Collocations for Students of English and corpus data (Collins Concordance Sampler). No partial points were given. Points were not deducted for misspelling of the acceptable collocates.
2. Vocabulary Treatment
learnt. Time constraints make it impossible to make them all the focus of explicit instruction. Therefore, it has been suggested that at advanced levels, vocabulary instruction should target less frequent senses of words and aim at creating new lexical networks (Lewis, 2000; McCarthy, 2007). Research showed that knowledge of collocations tends to lag behind learners’ vocabulary knowledge in general. German advanced learners, for example, were found to produce more than twice as many errors in translations of verb-noun collocations as in their translation of general lexical words (Bahns & Eldaw, 1993). Therefore, vocabulary treatment in this study focused on building the students’ knowledge of collocations, which was expected to ultimately help them make associations between the words and place them meaningfully within complex networks in the mental lexicon. As vocabulary learning takes place gradually (Nation, 2001), an effort was made to ensure that target words are encountered several times in different contexts.
The procedures adopted were as follows:
Step 1: Introduction to collocations (Week 1, time: 20 minutes)
The purpose of this stage was to raise students’ awareness of the importance of learning word collocations and give them some ideas how they can study collocations on their own. When they record the new words, the students were advised to write down the whole phrase from the text, not just the individual words. They were also introduced to two collocation dictionaries that were considered to be good reference sources - Oxford Collocations Dictionary and
LTP Dictionary of Selected Collocations. Finally, the students were directed to
useful Internet sources such as Corpus Concordance Sampler and guided through how to use them.
Step 2: Homework (Weeks 1-12, approximately 45 minutes per week)
Each homework assignment contained two vocabulary activities. In Activity One, the learners were presented with the list of target words, asked to check their meaning in a dictionary and copy at least one example of authentic usage for each word. The purpose of this activity was to familiarize learners with the form and the meaning of the target words, and encourage them to start keeping their own collocation notebooks.
The second activity was Collocation Crossword. The students were asked to complete a crossword, where the clues were the collocations that go with each of the target words. The clues were given in sets of four gap-filled sentences for each of the target words. The first example always came from the article that the learners were asked to read for homework. The other three examples were taken from the Oxford Collocations Dictionary, the Oxford Advanced Learner’s
Dictionary and the Corpus Concordance Sampler. The learners were warned that
in order to complete "the clue sentences", they sometimes might need to change the form of the word, but that they should use the basic dictionary form, i.e. the form given in the word list to complete the crossword. Typical collocates were highlighted in bold. An example of a set of clues is provided below.
Science has begun to the roots of child prodigies. Scientists are deeper and deeper into the secrets of the universe. The study was conducted to further into factors affecting concentration. The police were into his personal life. (*Target word: probe)
The crosswords were created with the use of a free software program,
This activity aimed at promoting the development of four different aspects of word knowledge: form, meaning, grammar and collocates. In order to fill in the gaps, the learners had to recall the meaning of the target word. As some sentences required different inflectional forms, they also had to think about grammatical properties. Highlighted collocates gave them typical examples of the usage of the target words. Finally, in order to complete the crossword, the students had to pay attention to spelling.
Vocabulary activities were followed by a set of comprehension questions related to the articles assigned for homework. Although no explicit attention was given to vocabulary at this stage, the homework readings provided additional exposure to the target words.
Step 3: Classwork (Weeks 2-13, approximately 45 minutes per week)
Classroom treatment of the target words consisted of three parts: homework review, completion of discussion questions where the target words were left out, and pairwork discussion conducted in English. Each part will now be discussed in more detail.
Homework review was done in two stages. First, the students were asked to share the examples of authentic usage they had found. This stage usually took about 10 minutes. Then the students were asked to compare their answers in Activity Two (crossword clues). After that, the model answers were provided and the students’ attention was drawn to the examples where the form of the word had to be changed. This stage also usually took about 10 minutes.
After the homework review was completed, the students were presented with a list of ten general discussion questions where the target words were blanked out.
For example, for ‘to forge’, the question was: In your opinion can giftedness be through environment by parents, schools and mentors? The students were asked to fill in the gaps and were then provided with the model answers. Usually, 10 minutes was spent on this activity.
Finally, the students were asked to work in pairs and have a discussion. During the discussion the students were looking at different pages. Each student had to ask five questions. The target words were again left out and their position was indicated with five slash lines. (Example: In your opinion can giftedness be /////
through environment by parents, schools and mentors?). The students were
instructed not to write anything but just to put the questions to their partner. The discussion usually took about 15 minutes.
The discussion preparation stage offered the students an opportunity to encounter the target word in a new context. The gap-filling exercise required the students to recall the meaning, spelling and grammatical properties of the word. The discussion that followed provided an additional focus on the spoken form. In addition to learning the pronunciation of individual words, drawing the students’ attention to collocations enabled them to learn the stress pattern of a phrase as a whole, leading to better intonation (Hill, 2000).
In summary, the vocabulary treatment ensured that the students were exposed to the target words at least eight times in seven different contexts, with an open-ended discussion offering potentially more encounters.
3. Post-test
The post-test consisted of a translation test and a collocation test. The tests were conducted in 12 sessions one week after the relevant vocabulary treatment.
The time allowed for each testing session was 10 minutes. The test followed the same scoring procedure as the pre-test.
RESULTS
1. Receptive vocabulary knowledge
The development of the students’ receptive vocabulary knowledge was examined by comparing their pre-test and post -test scores on the translation test.
a. Pre-test translation
For the analysis, the students’ scores were converted to percentages of the maximum possible score in each word class. This approach eliminated the effects of the differing numbers of words in the three word classes. The results indicate that most target words were new to the students. The participants were able to translate correctly about 33% of the target nouns, 24% of the verbs and 27% of the adjectives.
Table 1
Pre-test (translation) results: Descriptive statistics by grammatical class
Mean SD Range Nouns (n=37) 12.2 (32.9%) 5.1 4-26 (10.8% -70.2%) Verbs (n=18) 4.3 (23.9%) 3.3 0-14 (0% - 77.8%) Adjectives (n=23) 6.1 (26.5%) 3.5 2-15 (8.7% - 65.2%)
A close analysis of the students’ responses highlighted some of the problems L2 learners experience, such as difficulties with form recognition. The students’ responses supported Laufer’s (1990, 1997) findings that L2 learners tend to
confuse similar morphological forms. The mixing of similar orthographic forms was observed in all three grammatical classes. For example, physician was often translated as physicist, dumb was confused with dump, detain with maintain,
abundant with abandon, etc. Problems with form recognition made it sometimes
difficult to ascertain whether the students had any knowledge of the meaning of the target words.
b. Post-test translation
The results for all of the word classes demonstrate that the understanding of the words increased significantly after the vocabulary treatment. The students translated correctly 98% of the nouns, 91% of the verbs and 94% of the adjectives. The summary of the results is provided in Table 2.
Table 2
Post-test (translation) results: Descriptive statistics by grammatical class
Mean SD Range Nouns (n=37) 36.2 (97.8%) 1.9 29 – 37 (78.4% - 100%) Verbs (n=18) 16.3 (90.6%) 2.1 11 – 18 (61.1% - 100%) Adjectives (n=23) 21.6 (93.9%) 1.8 16 – 23 (69.6% - 100%) c. Pre-test-post-test differences
The Friedman test, the non-parametric equivalent of a repeated measures one-way analysis of variance, was carried out to assess the difference in pre- and post-test performance across the three grammatical classes. The results of the test indicate a significant difference between the six means: chi-square=86.1, with
df=5, p<.001. The Wilcoxon Signed Ranks tests were used to compare pre-test-post-test differences for each of the three grammatical classes. The differences were found to be statistically significant: for nouns z=3.73, p<.001, for verbs z =3.73 p<.001 and for adjectives z=3.74, p<.001.
2. Productive word knowledge a. Collocation pre-test
Productive knowledge of the target words was measured through a collocation test in which students were asked to provide three collocates for each target word. The maximum possible scores each student could get were 111 for nouns, 54 for verbs and 69 for adjectives. A summary of the results is provided in Table 3.
For all three grammatical classes, the scores on the collocation test fell significantly below those on the translation test. There were many words for which students were not able to produce more than one collocate, and not all of the produced word combinations were acceptable. Mean and Standard Deviation values do not indicate significant differences in students’ ability to produce collocations for the three word classes. The students were able to produce correctly 12.6% of the elicited collocations for nouns, 11.1% for verbs and 11.6 % for adjectives. Range values reveal some differences in the level of collocational knowledge of individual students. The widest spread of scores was observed for verbs, followed by nouns and then adjectives.
Table 3
Pre-test (collocations) results: Descriptive statistics by grammatical class
Mean SD Range Nouns (n=111) 14 (12.6%) 11.6 1-38 (1% -34.2%) Verbs (n=54) 6 (11.1%) 5.6 0-24 (0% - 44.4%) Adjectives (n=69) 8 (11.6%) 7.9 1-25 (1.4% - 36.23%)
Although overall the students performed better on the translation test, several instances were recorded where they were able to produce one or two correct collocations for the target words, even though they failed to provide correct translations. This phenomenon was observed in all three grammatical classes with similar frequency of occurrence: 1.6% for nouns, 1.4% for verbs and 1.8% for adjectives.
Acceptable responses were further examined in detail to establish possible patterns of associations for different word classes.
Nouns. A close analysis of students’ responses showed that just like in L1,
nouns tend to prompt nouns. Noun-noun patterns accounted for 62.7% of the collocations the students produced, 40.4% of which were the phrases where a collocate preceded the prompt word. A summary of the patterns identified is given in Table 4. (Words in bold print indicate the prompt word.)
Table 4
Pre-test collocation patterns for nouns
Pattern Frequency
noun+noun (e.g. organ broker) 40.7%
noun+noun (e.g. surrogate mother) 21% adjective+noun (e.g. technical prowess) 24.7%
verb+noun (e.g. issue edict) 8.6%
verb+preposition+noun (e.g. sue for malpractice) 0.9% verb+article+noun (e.g. obtain a patent) 0.9% noun+verb (e.g. eruption occurred) 0.9% noun+preposition (e.g. nomination for) 1.9%
preposition+noun (e.g. in uproar) 0.4%
In addition to noun-noun patterns, students produced a large number of adjective-noun collocations (24.7%). The number of verb-noun phrases was relatively low (8.6%). Four students also produced multiple word phrases (verb-preposition-noun and verb-article-noun), but they accounted for only a small number of responses. Other patterns included noun-verb, noun-preposition and preposition-noun phrases.
Verbs. When the target words were verbs, the following collocation patterns were identified:
Table 5
Pre-test collocation patterns for verbs
Pattern Frequency
verb+noun (e.g. impair ability) 77.8%
verb+article+noun (e.g. disclose a secret) 3% verb+ (article) + adjective+noun (e.g. sway public opinion) 2% verb+pronoun+preposition (e.g. indulge oneself in) 4% verb+adverb (e.g. forge further) 7.1% verb+preposition (e.g. convict for) 1.1% verb+preposition+noun (e.g. detain in custody) 1%
The data collected revealed that verbs are most likely to elicit nouns. In most cases nouns followed the verbs, although some verbs were found to produce nouns that would normally precede them in a sentence (e.g. prescribe-doctor,
probe-scientist, rumble-thunder etc.). The second most common pattern was
verb-adverb combinations, which accounted for about 7% of the responses. Some patterns were limited to certain verbs. For example, all verb-pronoun-preposition responses were prompted by indulge (e.g. indulge myself/oneself in).
Adjectives. Nouns were found to be the most common response to adjectives.
Adjective-noun combinations accounted for over 90% of the responses. A summary of the patterns identified is given in Table 6.
Table 6
Pre-test collocation patterns for adjectives
Pattern Frequency
adjective +noun (e.g. innate ability) 92.4% article+adjective +noun (e.g. an elaborate plan) 1.1% noun+adjective (e.g. accident prone) 0.4% verb+adjective +noun (e.g. prevent adverse effects) 0.9%
verb+adjective (e.g. play dumb) 1.8%
Adjectives also prompted some multiple word phrases (verb-adjective-noun). Some adjectives resulted in specific collocation patterns. For example, all recorded instances of verb-adjective patterns were either remain+rampant/vigilant or play+dumb. The phrase, prone+to, was the most common combination in adjective+preposition pattern, followed by notorious for. Prone was also the only adjective that prompted phrases where nouns preceded the adjective (e.g.
disease-prone, accident-prone).
b. Collocation post-test
A significant improvement in students’ knowledge of collocations was observed for all three grammatical classes. On average, students were able to provide 47% of elicited collocations for nouns, 42% for verbs and 44% for adjectives. A summary of the results is provided in Table 7.
Table 7
Post-test (collocations) results: Descriptive statistics by grammatical class
Mean SD Range Nouns (n=111) 52 (46.8%) 15 30-75 (27.0% - 67.6%) Verbs (n=54) 22.8 (42.2%) 7.5 12-39 (22.2% - 72.2%) Adjectives (n=69) 30.3 (44%) 9.7 18-49 (26.1% - 71.0%)
Data were further analyzed for collocation patterns within different grammatical classes.
association, followed by adjective-noun collocations. (See Table 8.)
Table 8
Post-test collocation patterns for nouns
Pattern Frequency
noun+noun (e.g. organ broker) 45.9%
noun+noun (e.g. surrogate mother) 13.4% adjective+noun (e.g. technical prowess) 15.9%
verb+noun (e.g. issue edict) 9.4%
verb+preposition+noun (e.g. sue for malpractice) 1.4% verb+article+noun (e.g. obtain a patent) 1.2% noun+verb (e.g. eruption occurred) 1.1% noun+preposition (e.g. nomination for) 7.4% noun+preposition+article+noun (e.g. eruption of a volcano) 2.9% noun+conjuction+noun (e.g. boom and bust) 0.8%
preposition+noun (e.g. in uproar) 0.6%
A moderate increase was observed in the number of verb-noun collocations and noun-preposition combinations (e.g. skirmish between, imbalance between/in,
ingredients for, etc.), two important patterns for the production of correct English
sentences. Verb-noun phrases, however, also contained a large number of errors. About 34% of the collocation errors in the noun class were linked to this pattern. Post-test data also included a complex pattern noun-preposition-article-noun (e.g.
eruption of a volcano, remnant of an era) and noun-conjunction-noun phrases (e.g. boom and bust) which were not observed in the pre-test phase.
Table 9
Post-test collocation patterns for verbs
Pattern Frequency
verb+noun (e.g. impair ability) 61.5%
noun+verb (e.g. doctor prescribed) 10.8%
verb+article+noun (e.g. disclose a secret) 3.9% verb+(article) + adjective+noun (e.g. sway public opinion) 1.4% verb+pronoun+preposition (e.g. indulge oneself in) 3.4%
verb+adverb (e.g. forge further) 4.7% verb+preposition (e.g. convict for) 10.9% verb+preposition+noun (e.g. detain in custody) 3.4%
Like in the pre-test, verbs were found to elicit mostly nouns. Most responses consisted of a single word. About 5% of verb-noun collocations also included an article, an adjective or both. While the number of verbal collocations with adverbs decreased, an increase of about 9% was observed in verb-preposition patterns. It is interesting to note that some of the unacceptable collocations for verbs resulted from the fact that learners would provide collocations for homophonous nouns. For example, for the target item to plot, some learners responded with phrases like
bomb-plot or make a plot rather than verb-node collocations such as plot secretly, plot against/with etc.
Adjectives. Adjectives formed the following collocation patterns:
Table 10
Post-test collocation patterns for adjectives
Pattern Frequency
article+adjective +noun (e.g. an elaborate plan) 2.7% noun+adjective (e.g. accident prone) 0.2% verb+adjective +noun (e.g. prevent adverse effects) 0.9%
verb+adjective (e.g. play dumb) 3.6%
adjective +preposition (e.g. notorious for) 2.8% adjective +preposition+noun (e.g. prone to accidents) 0.7%
Patterns for adjective collocations did not change significantly compared to the pre-test. Most adjectives were found to elicit single word noun responses. The number of verb-adjective and article-adjective-noun slightly increased, while the number of noun-adjective and adjective-preposition phrases decreased. No difference was observed in the verb-adjective-noun pattern.
c. Pre-test-post-test differences
The results of the Friedman test suggest that there are significant differences in the collocation pre-test - post-test scores across the three levels of word class: chi square 72.3, with df=5 p<.001. A pairwise comparison using the Wilcoxon signed rank test showed that the pre-test - post-test differences within grammatical classes were statistically significant. For the nouns, z=3.59, p<.001, for verbs z =3.64 p<.001 and for adjectives z=3.64, p<.001.
DISCUSSION
The study examined the effect of word class on the acquisition of L2 vocabulary. Data collected provides some insights into general developmental patterns of L2 lexical knowledge as well as the acquisition of collocational patterns for specific grammatical classes.
Marton, 1977; Bahns & Eldaw, 1993) that the knowledge of collocations develops more slowly than the knowledge of vocabulary in general. For all three grammatical classes examined, the scores on the collocation tests were significantly lower than the translation test scores. This means that, even when learners are able to comprehend the words, they may not always be able to use them correctly.
The results of the study also offer some new insights into the acquisition of different grammatical classes in L2. The prediction of the study was that, due to the conceptual differences and the different organization of the internal lexicon on definitions of nouns, verbs and adjectives, the three word classes would produce different behaviour. Nouns were expected to be learned more easily than verbs and verbs were expected to be learned more easily than adjectives.
The results of the study suggest that learners indeed have selective control of word classes - in other words, they do not treat nouns, verbs and adjectives equally. This was apparent from both the learners’ translations of the target words and the number of collocations they produced.
As predicted, nouns were easier to retain than verbs and adjectives. One reason for this may be that nouns tend to denote relatively simple entities in the mental lexicon, while verbs are always dependent on their noun arguments. Information is usually built up around nouns. Therefore, they play a pivotal role in the text (Woolard, 2000).
Contrary to initial predictions, however, adjectives were found to be learned more easily than verbs. On both the pre-test and the post-test understanding of the word classes was ranked noun > adjective > verb. These results differ from the
observed pattern of the lexical development of both native sp eakers of English and native speakers of Japanese. Although native speakers of Japanese were found to have a preference for verbs, that preference does not seem to be transferred to L2. The observed acquisition order differs from that of native speakers of English as well, who tend to acquire verbs earlier than adjectives.
One possible explanation for the results may be that the salience of nouns in English facilitates the indirect acquisition of words in their proximity, making the retention of adjectives easier. Adjectives may also have wider collocability than verbs, which may have allowed the students to produce a larger number of responses. Finally, it is possible that the acquisition of verbs is delayed due to semantic, syntactic and phonological complexities of this word class. Previous studies (e.g. Berman, 1988) showed that the items, structures and properties that are most universal and least marked across the languages are typically acquired earlier than formal features particular to specific languages. As discussed in the introduction, English verbs take a large number of inflections, which reduces their saliency. Furthermore, verbs may be more difficult to identify in connected speech as syllabic inflections tend to shorten verb duration (Davis et al., 1992). Difficulties result also from the specific stress pattern. While stressed syllables typically mark word onsets, bisyllabic English verbs normally stress the second syllable, making it difficult for learners to segment the speech stream (Marshall, 2003).
The analysis of the students’ responses within the three word classes revealed similarities in patterns of association with those of native speakers. Nouns were found to elicit nouns, verbs tended to generate nouns and adjectives typically resulted in noun responses.
Another trend observed was an increase in the number of multi-word phrases on the post-test in all the three word categories examined. As students were not explicitly advised to make longer phrases, this change in pattern suggests that explicit attention given to collocations naturally facilitates chunking - grouping of words in larger units.
A close examination of collocation errors indicates that Japanese students are most likely to have problems with verb-noun combinations. It is also interesting to note that students produced far more errors in verb-noun collocations when a noun was a node and a verb was a collocate, than the other way round.
Data collected also revealed that some collocation patterns may be under-represented in the students’ lexicon. No adverb-adjective collocations were observed and the number of verb-adverb phrases decreased on the post - test.
PEDAGOGICAL IMPLICATIONS
Data collected showed that learners’ knowledge of the meaning of L2 words far outstrips their knowledge of collocations and that more attention should be given to productive mastery of L2 vocabulary. Collocation-centered instruction is expected to facilitate the consolidation of words, increase the learners’ processing speed and enable them to access the words they recognize but cannot use effectively.
Collocations should not be taught randomly but as a part of a planned language input. Instructors should help learners to identify important language in the text, so that they can learn high-priority lexis. The advanced level presents a particular challenge because learners encounter many low-frequency items and extended and metaphorical senses of words, which makes sub-division and syllabus
organization difficult. Fortunately, with the help of powerful computer corpus linguistic techniques, it has become possible to identify both the covert patterns of lexical use in naturally occurring language and systematic irregularities in non-native output, which in their individual instances, may appear marginal or accidental. This allows teachers to design specific training programmes that would make up these deficiencies. Instructors should take into consideration both the corpus frequency of the phrases as well as the word class of the phrase constituents, so that they can select the language that is likely to be useful to the students but which they may overlook or have difficulties acquiring. Learners should be made aware of the more specific and the less overt collocational restrictions that are common in native speaker discourse. This is particularly important for advanced students who generally have mastered the basic rules and regularities of the language they are learning and often have more problems with usage than form.
Although it has sometimes been questioned whether explicit instruction helps L2 performance, there is plenty of evidence that advanced learners do resort to conscious knowledge in L2 production. Therefore, making learners aware of non-native tendencies (overused, underused or deviant patterns in their output) and giving them time and opportunity to practise should help them modify their behaviour in a more native-like direction.
The results of this study have shown that assumptions about the learners’ lexical development cannot be made based solely on lexical and morpho-syntactic features of their native tongue, or on data from native speakers of the target language. In the case of advanced Japanese learners, more attention should be paid to verbs and their collocations. In particular, more class time should be devoted to verb-noun collocations with a verb as a collocate and a noun as a node,
as this pattern seems to be a frequent source of difficulty for Japanese learners. The ability to access verbs is crucial for sentence building, and stronger links between verbs and other words in the lexicon are expected to promote fluency, reduce frustration and consequently increase students’ motivation.
Another word class that deserves more attention is adverbs, which seem to be under-represented in the vocabulary of Japanese learners. Adverbs are important both as a part of verbal collocations as well as adjective modifiers. Students are often encouraged to intensify adjectives in their writing to make it as expressive and colourful as possible. From a structural point of view, the use of intensifiers puts great demands on the learners’ collocational skills. Intensifiers are subject to a number of semantic, lexical and stylistic restrictions and include both restricted collocability (e.g. rancid butter, bitterly cold) and wide collocability (e.g. really
good/funny/necessary). Students tend to respond by either overusing all-round
adjective intensifiers such as really or very, resulting in clichéd expressions of little emotional force, or making too many points salient, giving the impression of wordiness and overstatement. As students’ intuitive judgment about the range of applicability and L2 restrictions, frequently does not conform to the native-speaker norm and register conventions, the result tend to be stylistically awkward sentences that give a strong impression of non-nativeness and typically create a feeling of unease in the reader (Lorenz, 1999).
The teaching of collocations should not be limited to two-word combinations. Instructors should try to direct learners’ attention to multi-word phrases as much as possible. Prefabricated language chunks play an important role in both acquiring and performing language. As Nattinger and DeCarrico (1992) observe, a great deal of language that people are exposed to every day is very routine and predictable. Chunking is known to reduce the processing load in comprehension
and consequently increases language processing speed (Mehnert, 1998; Morishita, 2008). While the output planning process is normally subconscious and highly automatic in L1 production, in the L2, limited exposure and lack of practice can hinder the successful passage of words from receptive to productive vocabulary (Laufer & Paribakht, 1998). Advanced learners are presented with an additional challenge, as they often encounter less frequent words which, by definition, require more processing effort (Wilson & Sperber, 1986). More mental effort means that less information can be stored at one time in short-term memory. As Nattinger and DeCarrico (1992) point out, such a reduction in storage capacity means that less linguistic data can be analyzed simultaneously, which interferes with both comprehension and language production.
Chunking can also reduce some of the errors in output. Longer memorized segments of speech allow learners to use expressions that they are not able to construct creatively from rules. For example, Japanese students are notoriously weak in the use of articles, which do not exist in their native language. Presenting the language in chunks, and having the students acquire new words in phrases, may help them to not only retain new vocabulary but also to produce grammatically correct utterances.
Finally, as vocabulary is an open system, it is impossible to complete the learning process in a typical institutional curriculum. Therefore, it is important to help learners develop the ability to select the words they are likely to need and the skills they need in order to build their knowledge of collocations and word grammar independently of the teacher.
FUTURE RESEARCH
First, more cross-linguistic research is necessary in order to assess the extent to which the acquisition of word classes depends on the input language. Second, more research should be conducted to examine whether similar collocation patterns would emerge with high frequency stimuli, and whether, and how, collocate types may change at lower proficiency levels. A comparison of native speakers’ and learners’ corpora could provide a further insight, not only into the number or patterns of collocations the two groups produce, but also the strength of these associations. More work will also be needed to determine whether there are any differences in the development of collocational knowledge across the subgroups of different grammatical classes. For example, a study conducted by De Bleser and Kauschke (2003) showed that German speakers find it easier to access intransitive verbs than transitive verbs. All verbs examined in this study can take an object and it would be interesting to see whether the number of verbal collocations would increase with stimuli limited to intransitive verbs. Finally, it would also be interesting to explore the nature and frequency of learners’ associations in free output, both written and spoken. For example, in a study conducted by Pine et al. (1996), checklist data revealed higher proportions of nouns than observational data. Adjectives presented in isolation, as they were in this study, are most likely to elicit nouns. Corpora analysis, however, has shown that adjective modification (e.g. adverb + adjective combinations) is more likely to occur in predicative than attributive positions. There is a possibility that free output may reveal new types of collocation patterns.
It is felt that if this study is able to make a small contribution towards a better understanding of non-native tendencies and the effect word class may have on the development of L2 knowledge, its pedagogical aims will have been achieved. It is hoped that the findings of this study will prompt instructors to pay more attention to potentially problematic word classes, and that this paper will stimulate further
research in L2 lexical acquisition. (7415 words)
REFERENCES
Bahns, J. and Eldaw, M. (1993). Should we teach EFL students collocations?
System, 21(1): 101-114.
Berman, R. (1988). Word class distinctions in developing grammars. In Y. Levy, I.M. Schlesinger and M.D.S. Braine (Eds.), Categories and processes in
language acquisition (pp. 45-72). Hillsdale, NJ: Lawrence Erlbaum.
Berman, R. and Slobin, D. I. (1994). Relating events in narrative: A
cross-linguistic developmental study. Hillsdale, NJ: Lawrence Erlbaum.
Choi, S. (1997). Language-specific input and early semantic development: evidence from children learning Korean. In D. Slobin (Ed.), The crosslinguistic
study of language acquisition, Vol. 5: Expanding the contexts. Hillsdale, NJ:
Lawrence Erlbaum.
Clancy, P.M. (1986). The acquisition of Japanese: The crosslinguistic study of
language acquisition, chapter 4. Hillsdale, NJ: Lawrence Erlbaum. Collins Concordance Sampler. [On-line], Available:
http://www.collins.co.uk/Corpus/CorpusSearch.aspx
Damasio, A.R. (1990). Category-related recognition defects as a clue to the neural substrates of knowledge. Trends in Neouroscience, 13: 95-98.
Davis, S., Morris, J. and Kelly, M.H. (1992). The causes of duration differences between English nouns and verbs. Unpublished manuscript.
De Bleser, R. and Kauschke, C. (2003). Acquisition and loss of nouns and verbs: parallel or divergent patterns? Journal of Neurolinguistics, 16:213-229. Goldfield, B.A. (1993). The noun bias in maternal speech to one-year-olds.
Journal of Child Language, 20:85-99.
Goldfield, B. A. (2000). Nouns before verbs in comprehension vs. production: the view from pragmatics. Journal of Child Language, 27: 501-520.
Haastrup, K. and Henriksen, B. (2000). Vocabulary acquisition: Acquiring depth of knowledge through network building. International Journal of Applied
Linguistics, 10: 221-240.
Hill, J. (2000). Revising priorities: from grammatical failure to collocational success. In Lewis, M. (Ed.) Teaching Collocation (pp. 47-69). Croatia: Thomson.
Hillis, A.E. and Caramazza, A. (1995). Representation of grammatical categories of words in the brain. Journal of Cognitive Neuoroscience, 7 (3): 396-407. Horst, M., Cobb, T., and Meara, P. (1998). Beyond a Clockwork Orange:
Acquiring second language vocabulary reading. Reading in a Foreign Language, 11: 207-223.
Kweon, S.O. and Kim H.R. (2008). Beyond raw frequency: Incidental vocabulary acquisition in extensive reading. Reading in a Foreign Language, 20 (2): 1-24. Laufer, B. (1990). Ease and difficulty in vocabulary learning: Some teaching
implications. Foreign Language Annals, 23 (2): 147-155.
Laufer, B. (1997). The lexical plight in second language reading. In J. Coady and T. Huckin (Eds.), Second language vocabulary acquisition (pp. 20-34). Cambridge: Cambridge University Press.
Laufer, B. and Paribakht, S. (1998). The relationship between passive and active vocabularies: Effects of language learning context. Language Learning, 48 (3): 365-391.
Lewis, M. (2000). There is nothing as practical as a good theory. In M. Lewis, (Ed.), Teaching collocation: Further developments in the lexical approach (pp. 10-27). Croatia: Heinle.
Lorenz, G.R. (1999). Adjective intensification - learners versus native speakers. Amsterdam: Rodopi.
LTP dictionary of selected collocations (1997). Croatia: Thomson.
developmental and acquired impairments. Journal of Neurolinguistics, 16: 67-84.
Marton, W. (1977). Foreign vocabulary learning as problem No.1 of language teaching at the advanced level. Interlanguage Studies Bulletin, 22: 33-57. McCarthy, M. (2007). Accessing development of advanced proficiency through
learner corpora. Center for Advanced Language Proficiency Education and Research, Position paper [On-line],
Available:
http://calper.la.psu.edu/downloads/pdfs/CALPER_ALP_Corpus.pdf
Meara, P. (1996). The dimensions of lexical competence. In G. Brown, K. Malmkjaer and J. Williams (Eds.), Performance and competence in second
language acquisition (pp. 35-53). UK: Cambridge University Press.
Meara, P. and Fitzpatrick, T. (2000). Lex30: an improved method of assessing productive vocabulary in an L2. System, 28 (1): 19-30.
Mehnert, U. (1998). The effects of different lengths of time for planning on second language performance. Studies in Second Language Acquisition, 20 (1): 83-108.
Miller, G. (1991). The science of words. New York: W.H. Freeman & Company. Morishita, M. (2008). The effect of different latencies on utterances of Japanese
EFL learners: A psycholinguistic study based on repeating and open question tasks. JACET Journal, 47: 17-33.
Nation, I.S. P. (2001). Teaching vocabulary in another language. Cambridge: Cambridge University Press.
Nattinger, J.R. and DeCarrico, J.S. (1992). Lexical Phrases and Language
Teaching. Oxford: Oxford University Press.
Oxford Advanced Learner’s Dictionary (1995). Oxford: Oxford University Press. Oxford Collocations Dictionary for Students of English (2002). Oxford: Oxford
Pigada, M. and Schmitt, R. (2006). Vocabulary acquisition from extensive reading: A case study. Reading in a Foreign Language, 18 (1): 1-25.
Pine, J. M., Lieven, E. V. M. and Rowland, C. F. (1996). Observational and checklist measures of vocabulary composition: What do they mean? Journal of
Child Language, 23: 573-589.
Polinsky, M. (2004). Word class distinctions in an incomplete grammar. In Diskin Ravid, D. and Bat-Zeev Shyldkrot, H. (Eds.), Perspectives on language and
language development (pp. 419-436). US: Kulwer Academic Press.
Read, J. (2000). Assessing vocabulary. Cambridge: Cambridge University Press. Schmitt, N. (1998). Measuring collocational knowledge : Key issues and an
experimental assessment procedure. I.T.L. 119-120: 27-47.
Tardif, T., Shatz, M. and Naigles, L. (1997). Caregiver speech and children’s use of nouns versus verbs: a comparison of English, Italian and Mandarin. Journal
of Child Language, 24: 535-565.
Wilson, D. and Sperber, D. (1986). Pragmatics and modularity. Papers from the
Parasession on Pragmatics and Grammatical Theory at the Twenty Second Regional Meeting. Chicago Linguistic Society, 22:67-84.
Woolard, G. (2000). Collocation - encouraging learner independence. In M. Lewis (Ed.), Teaching collocation (pp.28-46). Croatia: Heinle.
Zahar, R., Cobb, T. and Spada, N. (2001). Acquiring vocabulary through reading: Effects of frequency and contextual richness. The Canadian Modern Language
Appendix I-Target Words
Nouns (37) Verbs (18) Adjectives (23)
boom to convict abundant
broker to detain adverse
ceiling to disclose complacent
censorship to forge cordial
cynicism to impair counterfeit
edict to indulge dumb
eruption to materialize elaborate
gangrene to nurture fatal
grudge to outweigh genuine
imbalance to plot gruesome
ingredient to prescribe innate
latitude to p robe legitimate
loophole to reprimand notorious
malpractice to resent penal
nomination to scrutinize prone
nuance to slap prudent
patent to sway rampant
penalty to undermine rigorous
physician sensational premium sobering prosecution sophisticated prowess startling proxy vigilant raft
remnant seizure sin skeleton skirmish stigma surrogate technicality therapy trafficking transaction transplant uproar
Appendix II - A Sample of Vocabulary Pre-test / Post-test
Instructions: For each of the words on the list, write a Japanese translation. Then write 3 English words that can be used together with the words below.
Example: dog 犬 bark family friendly
If you don’t know the word, leave the line blank.
Target Words Translation Collocations (words that can be used together) boom
