Pietilä P., P. Lintunen & H.-M. Järvinen (toim.) 2006. Kielenoppija tänään – Language Learners of Today. AFinLAn vuosikirja 2006. Suomen soveltavan kielitieteen yhdistyksen julkaisuja no. 64. Jyväskylä. s. 13–36.
––––
EXPLORING THE QUALITY OF LEXICAL KNOWLEDGE IN THE LANGUAGE
LEARNER’S L1 AND L2
Birgit Henriksen University of Copenhagen
The dimensions of vocabulary size and quality of lexical knowledge have been described as central properties of the lexicon. Studies have shown that L2 vocabulary size is a strong predictor of L2 reading ability. Moreover, language learners below a certain L2 vocabulary threshold are not able to transfer effective L1 processing skills to an L2 task. Strong correlations also exist between quality measures of lexical knowledge and reading and listening. Quality of lexical knowledge has been defined in many ways, both as increased understanding of word meaning, as comprehensive knowledge about the word and its use and as developing interrelationships between the words in the mental lexicon. The paper will provide a review of research within the field, exemplifying with data from a cross-sectional study of foreign language learners’ lexical knowledge in both L1 (Danish) and L2 (English).
Keywords: lexical knowledge, mental lexicon, network knowledge
1 INTRODUCTION
We have witnessed a shift of attention to lexical competence and an increase in lexical studies in second language acquisition research over the last two decades. Despite an array of vocabulary research
studies and an upsurge in practical suggestions for working with vocabulary in the language, we are, however, still trying to get to grips with our basic understanding of the construct of vocabulary knowledge and the relationship between lexical competence and different language skills. Moreover, we are also struggling to find ways of developing valid and reliable tools for measuring different aspects of lexical competence and for tracking vocabulary development. The primary focus in this paper will be the construct of lexical quality and here especially the question of measuring degrees of network knowledge.The following issues will be addressed 1) How should we define lexical competence? 2) What is the relationship between lexical competence and the different language skills? 3) Review of research on network knowledge 4) Examples from a cross-sectional study of foreign language learners’
network knowledge in both their L1 and their L2 5) Research perspectives and pedagogical implications of focusing on network knowledge.
2 DEFINING LEXICAL COMPETENCE
The dimensions of vocabulary size and depth or quality of lexical knowledge have been described as central properties of the mental lexicon, both in our L1 and our L2. Vocabulary size views lexical knowledge from a purely quantitative perspective and clearly refers to the number of words in our mental store. Few would disagree with the old truism that “the MORE – the better”. When we are talking about somebody having “a good vocabulary”, the construct of size is often what is referred to, and different test procedures for measuring vocabulary size have been developed: word recognition tasks, translation tasks, picture matching tasks, definition tasks, different interview procedures and written production measures.
Some are receptive and some are productive; some look at words in isolation, while others use a more contextualized approach in which the words are presented in a contextual frame, often in the
form of a short sentence. It is, however, not surprising that the reports we get on our learners’ vocabulary size will be highly dependent on the test format that we choose to use. The different measures reflect different underlying conceptions of lexical knowledge and tap different stages of vocabulary acquisition ranging from mere recognition of word form across extensive knowledge of word meaning to productive use ability.
Another way of looking at “more” equals “better” is to look at vocabulary knowledge from a more qualitative perspective and define “more” in terms of the type and the quality of knowledge that the learner has about the individual words in the mental lexicon.
Different approaches to the notion of depth or quality of lexical knowledge have been proposed (e.g. Henriksen 1999). Here, I will lean on the most recent definition outlined by Read (2004), who tries to disentangle the various ways the term ‘depth of vocabulary knowledge’ has been employed in the research literature. According to Read, the concept of depth or quality of lexical knowledge has been used by researchers in three distinct ways: 1) precision of meaning, 2) comprehensive word knowledge and 3) network knowledge.
Precision of meaning is defined as “elaborated and specific knowledge of word meaning” (Read 2004: 211). This notion rests on the underlying assumption that during the first encounters with a word, learners often develop a vague understanding of what a word means, i.e. what the lexical item refers to. In more naturalistic approaches to language learning, it is assumed that precision of meaning will come gradually during subsequent encounters with the same word in varying contexts, often supported by the learners’
use of different lexical inferencing procedures, i.e. lexical guessing strategies. Henriksen (1999) has described this process of vocab- ulary acquisition as a gradual or cumulative development along a partial to precise knowledge dimension. Dictionary use, glossing of texts or different definitional techniques are used to support and speed up this process of mapping meaning on to form and of acquiring precision of word meaning.
Comprehensive word knowledge is defined by Read as
“knowledge of a word which includes not only its semantic features but also its orthographic, phonological, morphological, syntactic, collocational and pragmatic characteristics” (Read 2004: 211). Read here refers to all the other aspects of word knowledge which go beyond the mere semantic content which was alluded to in the first dimension, which, as we have heard, was concerned with the propositional content of a lexical item. This second dimension of
‘depth of lexical knowledge’ echoes the comprehensive lists of both receptive and productive word knowledge which have been outlined in the research literature (Richards 1976; Nation 1990, 2001). Apart from the knowledge of written and spoken word form, a number of these knowledge components refer to the learners’ knowledge of the use restrictions connected with a specific lexical item.
The third use of the term ‘depth of lexical knowledge’ refers to network knowledge. Quality of lexical knowledge is here defined by Read as “the incorporation of the word into a lexical network in the mental lexicon, together with the ability to link it – and distinguish it from – related words” (Read 2004: 212). As pointed out by a number of researchers (Ringbom 1983; Aitchison 1987;
Meara1996; Henriksen 1996, 1999), developing lexical competence not only involves the process of item-learning, i.e. adding new words to the lexical store, but also includes the important process of creating links between the lexical items in the mental lexicon.
The creation and strengthening of numerous links between the items and the restructuring of the relations between the lexical items, i.e.
network building, is a continuous process which takes place all the time both as the vocabulary expands in terms of size and as our knowledge of the individual items develops.
Although Meara (1996) and his co-researchers (Wilks & Meara 2001; Meara & Wolter 2004) deal with the same notion of linkage between lexical items in the mental lexicon, they do not regard network knowledge as a feature of the individual lexical items, but prefer to look at the overall organisational structure of the mental lexicon. According to their research paradigm, lexical organisation
refers to the number of links, i.e. the density of the lexical net and not to the quality or types of links created between the individual lexical items. Meara (1996) has argued that the notion of vocabulary size is crucial if we look at language beginners, but he stresses the increasing importance of a well-organized lexicon for more advanced learners. As the learner expands his vocabulary in terms of size, the organisational properties of the lexicon become more and more important in order to ensure smooth and effective access to the larger and larger mental store of lexical items.
3 LEXICAL COMPETENCE AND LANGUAGE SKILLS?
A central question is how much and which type of lexical knowledge a language learner needs to be able to use a language effectively, or phrased differently: what is the relationship between lexical competence and the different language skills? Nation and Waring (1997) have approached this question by looking at text coverage.
An analysis of a corpus of written English texts showed that 80%
of the words in any of these texts belonged to the first 2000 most frequent words in English. As can be seen from Figure 1 below, a number of researchers have looked at text coverage in relation to different text types and language skills.
Studies Mode and Vocabulary size Text
Text types coverage
Nation and Waring (1997) Written texts First 2000 words 80 % Hu and Nation (2000) Conversation First 2000 words 90 % Fiction First 2000 words 90 % Academic texts First 2000 words 75 % Adolphs and Schmitt (2003) Spoken discourse First 2000 words 95 % FIGURE 1. Text coverage and vocabulary size.
It is evident from these approximate figures of text coverage that a learner can get far with knowledge of the most frequent, basic words in the vocabulary. Text coverage, however, does not tell us about the number of words or the quality of the lexical knowledge, which is needed to perform adequately and efficiently in a real receptive or productive task. Learners probably need a threshold vocabulary between 3000 and 5000 words to be able to read effortlessly in a foreign language and to achieve adequate text comprehension.
Vocabulary sizes ranging from 5000 to 10,000 and even up to 15,000 have been mentioned as a prerequisite for reading and understanding more advanced texts, for example in relation to academic studies.
Laufer (1997) has stressed that the learner not only needs a vocabulary of a certain size, but must also be able to access these words smoothly and effortlessly, i.e. the learners’ knowledge of the words must be of a certain quality to ensure efficient language use. The term “sight vocabulary” is used to refer to lexical items which can be accessed immediately, i.e. “on sight” by the learner.
The ability to recognize and decode a large number of words in a text without mental effort will free cognitive processing capacity for other central aspects of the comprehension process. Moreover, it has been shown that learners below a certain threshold vocabulary are not able to transfer effective reading procedures which they may have developed in their first language to the L2 task.
Studies from L1 research have stressed the role of vocabulary knowledge in reading comprehension. As can be seen from figure 2, a number of L2 studies have also found varying but still statistically significant intercorrelations between L2 vocabulary size and L2 reading, as well as listening comprehension.
Studies Lexical Relational Listening or Correlation construct aspects reading
comprehension
Laufer 1992, 1996 size reading .50 and .75
Bonk 2000 lexical listening .45
familiarity
Pemberton 2002 size listening .73
Henriksen, Albrechtsen
& Haastrup 2004 size reading .79 and .85
Qian 1998/1999 size & depth synonymy reading .78 and .82 and
collocation
Mecartty 2000 depth synonymy reading .50 and .38
and listening
Qian 2002 size & depth synonymy reading .74 and .77 and
collocation
Stæhr Jensen 2005 size & depth synonymy listening . 68 and .72 and
collocation
FIGURE 2. Correlations between vocabulary knowledge and reading and/
or listening comprehension.
Ringbom (1983) has argued convincingly that vocabulary acquisition is not only a process of item-learning but also of system- building. Lexical items are not stored randomly in the mental lexicon but are linked together in various ways which provide a structure or a system which will give the language user smoother and more efficient access to the items in the process of language use. The correlational evidence from the studies outlined in Figure 2 indicates that a strong relationship between two measures of network knowledge, namely synonymy and collocational knowledge, and reading and listening exists. The empirical studies mentioned here have only looked into the relationship between receptive skills and the different vocabulary measures. It is likely that the same close relationship between lexical competence and productive skills can also be hypothesized.
4 RESEARCH ON NETWORK KNOWLEDGE
As can be seen from the correlational studies in Figure 2, depth is here primarily associated with Read’s third dimension, network knowledge. In the following, I will focus on some of the research which has been carried out on network knowledge with an emphasis on the types of studies that have influenced and informed the procedures used in my own empirical research.
4.1 PRODUCTIVE WORD ASSOCIATION TASKS
Many researchers have used the traditional word association technique to describe the organisational properties of the lexicon.
In this task type, the informants are asked to write down or say the first word which comes to mind when they read or hear a stimulus word. The structure of the lexical network is then categorized by looking at the different response types given by the informant (see Figure 3).
Semantically related response Stimulus Form-related Syntagmatic Paradigmatic Canonical
response
black back bird colour night
hand land shake nail finger
FIGURE 3. Different response types in word association research.
Studies have shown that both young children speaking their mother tongue and L2 beginners tend to give a form-related or clang response whereas older and more experienced learners will predominantly supply a semantically related response. This has been interpreted as an indication that form relations play a more dominant role in the initial stages of processing and learning, whereas semantic factors play an increasing role as the mental lexicon develops. If we look more closely at the semantically related links, many word association studies have shown that learners at the earlier
stages overwhelmingly supply a syntagmatic link, i.e. a collo- cational structure (the word ‘bird’ as a response to ‘black’ or the word ‘shake’ as a response to ‘hand’). Later the paradigmatic links predominate, i.e. more hierarchical responses like hyponyms, parts to whole, synonyms and antonyms (the word ‘colour’ as a response to ‘black’ and ‘nail’ as a response to ‘hand’). Other researchers have shown that the response patterns of very advanced learners, for example adult L1 speakers, reflect a shift back to more syntagmatic responses, triggered by an increase in precision and comprehensiveness of word knowledge. It is important to note, that the L1 study by Deese (1965) and the L1/L2 study by Nissen (2002) have documented that the response patterns found are highly affected by the word classes tested. Namei (2002; 2004), who has worked with cross-sectional data from a large number of bilingual informants, not only looked at semantic properties of the responses, but analysed her data in relation to word frequency. Interestingly enough, she found an increased tendency across time to respond with low frequent, semantically related words.
Based on research findings (Entwisle 1966; Söderman 1989;
Wolter 2001; Namei 2002, 2004), one could, however, argue that the shifting role of form and meaning is not primarily a feature of the whole lexicon, but is more related to the learner’s level of knowledge of the individual lexical item in the mental lexicon. In the initial phases of learning, most of the lexical items are relatively unknown and, consequently, we find that formal factors seem to play a fairly significant role, but with increased proficiency the words we know well become more and more meaning driven. When we are dealing with newer, more unfamiliar vocabulary items, we are still inclined to rely more on the form-driven, phonological information in the lexical entry. The same has been found for the relative distribution of the semantically related responses. When a learner is dealing with a new, perhaps low frequent stimulus word, the response tends first to be phonologically related, then syntag- matically based, whereas the learner with increased knowledge of
the same lexical item is more likely to supply a paradigmatic response.
L1 word association research has also shown that many native speakers tend to give the same responses to the same stimulus words.
Words like ‘ice’ and ‘hot’ are, not surprisingly, very frequently given as responses to the word ‘cold’. These more stable responses can therefore be characterized as more canonical or central in the lexical net than other more infrequently given responses that are likely to be more context-dependent. It could be hypothesized that the canonical links (for example the link between ‘black’ and ‘night’
and ‘hand’ and ‘finger’), play a central role in the structuring of the mental lexicon. Canonical lexical items may perhaps function as bridges or pointers between the different parts of the lexical net.
Schmitt (1998) has addressed the question of using these native- like response types as an indicator of development of network knowledge, suggesting that we should give higher scores to L2 learner associations which mirror the canonical associations given most frequently by native speakers. In this line of research, quality of organisational structure or network knowledge is therefore defined on the basis of canonicity or prototypicality of response types. Namei (2002) for example found an increased tendency to give more typical responses as a function of degree of word knowledge.
4.2 RECEPTIVE NETWORK TASKS
The word association research reviewed so far has been based on a productive version of the word association task. Read has developed a receptive version of the word association task. In the first version of his word associates task (Read 1993), which is shown in Figure 4, the informants were asked to identify four possible links to the stimulus words. The relations can either be syntagmatic, paradigmatic or analytic. The four other words are distractors, i.e.
they are unrelated to the stimulus word. The test format was later
changed somewhat by Read (1998, 2000) to minimize the guessing element.
Edit
arithmetic film pole publishing
revise risk surface text
FIGURE 4. The original version of the word associates test.
The word associates test has been developed to find a reliable task to measure learners’ network knowledge and the format has been further explored and developed by numerous researchers (for example Greidanus & Nienhuis, 2001; Greidanus et al. 2004) and the word associates format has already been used in a number of studies to assess L2 network knowledge (Qian 1999, 2002).
A number of studies have looked at the difference between network knowledge in the L1 and the L2. Some of the research which has used the productive word association task has already been reviewed. Let us now turn to comparisons of receptive network knowledge in the L1 and the L2.
Three studies (Schoonen and Verhallen 1998; Verhallen et al.
1999; Greidanus et al. 2005) used new revised versions of the Read word associates task. The network density task used by Wilks and Meara (2002) and Skriver (2005) is, however, a different test format in which the informants were asked to identify possible relational links in random chains of words.
All these studies showed that native speakers were able to identify a significantly higher proportion of network links than L2 learners. Moreover, a progression in network knowledge was found when they compared L2 learners at different university levels. In other words, the results indicate that the language learner’s mental lexicon is less dense than that of the native speaker. An implication of this may be that the learners’ retrieval paths are fewer and longer;
a fact which may well affect learners’ abilities to access words
effectively in L2 reception. The results, however, also showed that a progression towards native speaker response behaviour does take place across time.
5 A STUDY COMPARING L1 AND L2 NETWORK KNOWLEDGE
I will now report on a cross-sectional study of network knowledge that I have been involved in in the last few years. The research has been highly inspired by previous network studies that have used the productive and the receptive version of the word association task. Most of these studies have compared L2 learners with native speaker controls. Unfortunately, only a few researchers, for example Namei (2002) and Skriver (2005), have collected data from the same learners in both their L1 and their L2. Our study also includes L1 (Danish) and L2 (English) data from the same informants.
The empirical data has been collected as part of the Danish research project “Processes in writing and vocabulary acquisition in English as a foreign language” which I am working on at the moment together with Dorte Albrechtsen from the University of Copenhagen and Kirsten Haastrup from the Copenhagen Business School (see Albrechtsen et al. 2004; Haastrup et al. 2004; Henrik- sen et al. 2004). The overall aim of the project has been to investigate the relationship between knowledge and skills manifested by the same informants in the foreign language and the mother tongue.
Data was collected both over time (longitudinally) and across three learner groups: grade 7 (aged 13-14), grade 10 (1. year of high school, aged 16-17) and grade 13 (1. year of university, mean age 20). A range of tasks were used: two different types of writing tasks, lexical inferencing tasks, in which the informants had to guess the meaning of unknown words when reading, and three different lexical network tasks. Moreover, the informants were given a vocabulary size test and a reading test in both languages. By giving
our informants such a wide range of tasks, it has been our hope that it will be possible to set up complex profiles of our informants’
abilities within these different areas in their L1 and L2. Moreover, we will explore the possible relationship between the various skills and lexical knowledge.
In this paper, I will only focus on the lexical network data collected at time 1. In the network sub-study of our project, the general research aim has been to find ways of detecting and describing possible differences in the relational links between lexical items across the three educational groups and across the two languages. We are interested in finding ways of tapping the learners’ network knowledge on the assumption that not only vocabulary size, but also the structural qualities of our informants’
lexical knowledge are important for smooth and efficient language production, for example when they are carrying out writing tasks.
For reception, too, it is assumed that not only vocabulary size, but also network knowledge is important for the learners, when they are trying to guess the meaning of unknown words in our lexical inferencing tasks.
5.1 THE TWO NETWORK TASKS USED AT TIME 1
Two types of network tasks were used at time 1: a traditional productive word association task and a receptive word connection task that was developed specifically for this project. For both tasks, the Danish task was a translated version of the English task. Baseline data from 127 native speakers of English and 108 native speakers of Danish, carrying out the same pen-and-paper tasks, has also been collected. The informants in both norming groups are students in their twenties. A comparison between the informant data and the baseline data is expected to reveal a gradual approximation from the lowest to the highest learner group to native speaker response patterns.
Our informants from the three educational levels were first asked to perform the two tasks as pen-and-paper tasks without concurrent think-aloud. After the completion of each task, retrospective data, during which the informants explained and qualified the choices made in the pen-and-paper task, was collected.
The think-aloud data has been used to support the analyses and scoring procedures. The tasks were counterbalanced as to language, and data was collected with a week’s interval between the Danish and English versions of the same task.
In the word association task, 48 high-frequency nouns and adjectives (see Figure 5) were read out to the informants at 15- second intervals. When a word was read out, the informants had to write down the first two words that came to mind in response to the stimulus word.
Nouns Adjectives
moon, child, fruit, house, woman, cold, beautiful, afraid, hungry, slow, chair, hand, bread, head, spider, sweet, dark, deep, soft, short, quiet river, lion, eagle, ocean, soldier, bitter, yellow, long, high, hard, blue, butter, window, sheep, bed, thirsty, white, black, red, sour, heavy, stomach, cheese, mountain, doctor, green
foot
FIGURE 5. Stimulus words included in the word association task.
In the word connection task, the first 24 words that were included in the word association task were given again to the informants.
COLD: war water frost hand hot warm snow pain winter ice FIGURE 6. An example item from the word connection task.
Each of the target words was presented with 10 words that could be expected to be linked to the target word in the mental lexicon.
Five out of the 10 potential links, represent the five most frequent canonical associations given by native speakers in various word
association norming lists. The five other words have also been taken from these native speaker word association norming lists. They are semantically related but more infrequent responses which were given by only one native speaker in the norming data. Unlike the task format developed by Read, all words in our format represent potential but canonical vs. peripheral links in the lexical net. Our informants were instructed to choose the five words that they found were most strongly connected with the target word. Time on task was 20 minutes.
5.2 HOW THE DATA WAS ANALYSED AND SCORED
On the basis of the results from our own native speaker norming groups, strong links in the English and the Danish word connection task were identified (GB =126 strong links; DK = 120 strong links).
A score of 1 was awarded for each correctly identified strong link.
The results from the two languages are reported in percentages of the maximum possible scores for each task.
A number of severe scoring problems have been discussed by researchers working with the productive word association task. One of the major problems has been the question of distinguishing clearly enough between paradigmatic and syntagmatic responses.
Moreover, it has been difficult to distinguish early syntagmatic and late syntagmatic responses. Finally, the distinction between idiosyncratic, that is, personal and often peculiar responses, and semantically related links has often been based on highly subjective criteria and has therefore threatened the reliability of the scoring procedure. A new procedure for analysing the word association data was therefore developed for this study.
On the basis of the results from our two native speaker norming groups, the most frequent, i.e. the canonical word association responses for the English and the Danish data, were identified. Each of the 96 responses given by the informants was then categorized and awarded a separate response type score according to the taxonomy of response types developed for the study. This procedure
is outlined in Figure 7 with an example from the L2 stimulus word
‘bread’.
Response type Example with the stimulus Score word ‘bread’
Inability to supply an L1 or L2 brød (L1 translation) 0
response bread (repetition of stimulus)
Form-related response red 1
High-frequent, non-canonical white 2
semantically related response
High-frequent canonical response food 3
Low-frequent canonical response loaf 4
Low-frequent, non-canonical but grainy 5
semantically related response
FIGURE 7. Scoring procedure for the word association task.
An overall response type score was calculated for each informant.
Not surprisingly, some of the weaker informants often gave the same response to a number of different stimulus words. To minimize the effect of such mere repetition across the data set, the response type score was then multiplied with a lexical variation score which had been calculated on the basis of the informant’s 96 responses.
This scoring procedure generated an overall word association score for each informant in both their L1 and their L2. This overall score is seen as a measure of the individual learner’s productive network knowledge for high frequent nouns and adjectives. However, it must be pointed out, that the score can therefore only give an estimate of network knowledge for a certain stratum of the lexicon.
5.2.1 Results
Our general predictions were that we would find a statistically significant difference in the results across the two languages for the informants from grade 7 and for a number of informants from grade 10. Since we are dealing with very high frequent nouns and adjectives in both network tasks, it could, however, be expected
that the more advanced students, especially in grade 13 would do equally well in both languages. It was also our prediction that the results from the two tasks could differ markedly. The word connection task is receptive, and the informants are given ample time for reflection, whereas the word association task is productive, and the informants are given only 15 seconds to write down the two responses to each stimulus word.
The mean scores for the three educational levels across the two languages for the two network tasks are reported in Table 1.
TABLE 1. Comparison between the L1 and L2 on the two network tasks.
GR 7 GR 10 GR 13 GR 7 GR 10 GR 13
WCT WCT WCT WCT WCT WCT
n= 29 n= 29 n= 29 n= 29 n= 29 n= 29
GB M64.89% M71.29% M70.60% M151.83 M208.65 M221.77 sd28.95 sd4.6 sd5.36 sd49.76 sd25.26 sd28.95 DK M71.09% M74.83% M 74.46% M217.89 M238.65 M239.17
sd 6.89 sd3.55 sd5.04 sd 24.23 sd19.82 sd31.85
Let us first look at the results from the receptive word connection task reported as WCT in table 1. To determine whether there was an overall difference between the two languages on the WCT scores from the three educational levels, an analysis of variance (ANO- VA) was performed. The ANOVA showed an overall statistically significant difference between the English and the Danish data (F (1, 84) = 62.308, p < .001). The result from the overall ANOVA was confirmed by the ANOVA for the informants for all three grade levels (p < .001 (grade 7: F (1, 28) = 22.963, p<.001; grade 10: F (1, 28) = 18.187, p<.001; grade 13: F (1, 28) = 24.674, p<.001).
The same statistical procedure was used on the productive word association data, reported as WAT in table 1. The overall ANOVA realized a statistically significant difference (F (1, 84) = 111.527, p
< .001). The ANOVA for all three educational levels showed significant language differences, i.e. the DK scores were higher than the GB scores for all three informant groups (grade 7: F (1,
28) = 56.282, p<.001; grade 10: F (1, 28) = 37.065, p < .001; grade 13: F (1, 28) = 22.659, p <.001). A qualitative analysis of the response types reveals that the learners from all three educational levels supplied significantly fewer canonical and low frequent links in the L2 task compared with the L1 task. For the grade 7 and grade 10 informants, a significant difference is also found in the lexical variation score across the two languages. In other words, these learners were more prone to repeat the same response words when they were doing the English task.
The results from the two network tasks are somewhat surprising. As mentioned earlier, it was expected that at least the university informants would have done equally well in the two languages on both tasks. What is more surprising is that a language difference could be found for both the receptive and the productive task. It is important to point out again that all the words included in the two tasks are very high-frequent nouns and adjectives, as can be seen from the list in Figure 5. We know that most of these words are known to all the informants from all three educational levels. If we had devised a vocabulary size test tapping their knowledge of word meaning on these specific lexical items, it could therefore be expected that our informants would have scored equally well both across the educational levels and across the two languages. But the network tasks reveal that even our high level informants do not display the same network knowledge in English as in Danish on these very common, high frequent words. It must again be pointed out, that the results on both tasks can only give us an estimate for network knowledge for a certain stratum of the lexicon. If we had tapped into the informants’ knowledge of high-frequent verbs or more infrequent lexical items from all the major word classes, it is highly likely that the differences in quality of lexical knowledge between the two languages would have been even more pronounced.
6 RESEARCH PERSPECTIVES AND PEDAGOGICAL IMPLICATIONS
In our network study we have looked at receptive as well as productive network knowledge of very high frequent vocabulary items. The results have shown that L2 learners lag behind in their L2 network knowledge compared to their L1 knowledge, both receptively and productively. What is surprising, however, is the fact that even the high level learners do significantly better in their L1 compared to the L2 in tasks which tap into the knowledge of these extremely high frequent words. It is highly probable that they have been exposed to these lexical items many times before also in the target language. Many of the words included are even close cognates to the same lexical items in their L1. Our results confirm the point that has been made by Aitchison (1987) about the slow, almost tortoise like development of network knowledge. It takes a long time and a very large amount of language exposure to establish the quantity and types of network links found in the native speaker lexicon.
Again it is important to stress that we have only been tapping into the network knowledge of a limited number of high frequent words from two word classes, so we are only opening a small door into the mental store. As noted above, it could be expected that the differences found between L1 and L2 would have been even more pronounced, if we had tested a much wider range of vocabulary items. Or to phrase it differently: even our very advanced learners seem to have weaknesses in their network knowledge on the limited range of high frequent vocabulary which was included in our study.
We may perhaps have cause to question the overall quality of their lexical competence, if we solely base our evaluation of their vocabulary on a traditional size measure.
If we believe that the development of network knowledge is an important factor for smooth and efficient language use, as indicated by the correlational evidence presented in Figure 2, we
must find different ways of supplementing tests of vocabulary size with quality measures of lexical competence. It is important that we acknowledge the complexity of lexical competence and the cumulative, i.e. the step by step nature of vocabulary acquisition.
The line of research instigated by John Read’s and Paul Meara’s work is a valuable step in this direction, but we must remember that their test formats only focus on one of the three dimensions of lexical quality outlined in this paper. More work on and attention to the development of sensitive test instruments for other aspects of lexical competence would give researchers as well as practicing teachers more accurate tools to describe and measure the extremely complex and incremental lexical progress of our learners. It is my own hope, in a future research programme, to be able to develop a procedure for tapping network knowledge in line with the approach used in our own study, but based on the lexical items tested in for example Nation’s widely used vocabulary size test. This would make it possible to combine a measure of network knowledge across a range of frequency bands and word classes with a standardized measure of vocabulary size. Such a procedure would give us a more multidimensional estimate of a language learner ’s lexical competence.
If we look at the construct of lexical quality from a pedagogical perspective, the results from the research on network knowledge reviewed here seem to indicate a need to focus more specifically on procedures that will support the development of linkages in the lexical net. Namei (2004) has argued that the lack of sufficient lexical knowledge which she detected in her informants’ response behaviour, may be due to limited exposure to the target language, and she therefore calls for pedagogical approaches which allow for extended exposure to lots and lots of varied and authentic target language input inside and outside the language classroom. The research by Paribakht and Wesche (1996; 1997) and Joe (1995) has shown the positive effect of combining a reading approach with text-related vocabulary exercises. If we look at more explicit
teaching procedures that advocate the use of lexical tasks (Newton 1995), it may be the case that different task types, for example sorting tasks and gradation tasks (Haastrup & Henriksen 2000) and the use of mind maps that operate with words in lexical fields and which assist the creation of linkages between words, may be useful didactic tools. Contextual and definitional techniques which represent the lexical items in a contextual frame rich in relational links may also prove to be a useful path to follow.
REFERENCES
Adolphs, S. & N. Schmitt 2004. Vocabulary coverage according to spoken discourse context. In P. Bogaards & B. Laufer (eds.) Vocabulary in a Second Language. Amsterdam: John Benjamins, 39–49.
Aitchison, J. 1987. Words in the Mind: An Introduction to the Mental Lexicon. Oxford: Blackwell.
Albrechtsen, D., K. Haastrup & B. Henriksen 2004. Attention to
argumentation in learner text production. How do we capture learner ability in argumentation? Angles on the English-speaking World 4, 111–128.
Bonk, W. 2000. Second language lexical knowledge and listening comprehension. International Journal of Listening 14, 14–31.
Deese, J. 1965. The Structure of Associations in Language and Thought.
Baltimore: The John Hopkins University Press.
Entwisle, D. R. 1966. Word Associations of Young Children. Baltimore:
The Johns Hopkins University Press.
Greidanus, T., P. Bogaards, E. van der Linden, L. Nienhuis & T. de Wolf 2004. The construction and validation of a deep word knowledge test for advanced learners of French. In P. Bogaards & B. Laufer (eds.) Vocabulary in a Second Language. Amsterdam: John Benjamins, 191–208.
Greidanus, T. & L. Nienhuis 2001. Testing the quality of word knowledge in L2 by means of word associations: types of distracters and types of associations. The Modern Language Journal 85, 567–577.
Greidanus, T., B. Beks & R. Wakely 2005. Testing the development of French word knowledge by advanced Dutch- and English-speaking learners and native speakers. The Modern Language Journal 89: 2, 221–238.
Haastrup, K. & B. Henriksen 2000. Vocabulary acquisition: acquiring depth of knowledge through network building. International Journal of Applied Linguistics 10, 221–239.
Haastrup, K., D. Albrechtsen & B. Henriksen 2004. Lexical inferencing processes in L1 and L2. Same or different? Focus on issues in design and method. Angles on the English-speaking World 4, 111–128.
Henriksen, B. 1999. Three dimensions of vocabulary development. Studies in Second Language Acquisition 21, 303–317.
Henriksen, B., D. Albrechtsen & K. Haastrup 2004. The relationship between vocabulary size and reading comprehension in the L2.
Angles on the English-speaking World 4, 129–140.
Hu, M. and P. Nation 2000. Unknown vocabulary density and reading comprehension. Reading in a Foreign Language 13, 403–430.
Joe, A. 1995. Text-based tasks and incidental vocabulary learning. Second Language Research 11, 149–158.
Laufer, B. 1992. How much lexis is necessary for reading comprehension?
In P. J. L. Arnaud & H. Béjoint (eds.) Vocabulary and Applied Linguistics. London: Macmillan, 126–132.
Laufer, B. 1997. What’s in a word that makes it hard or easy: some intralexical factors that affect the learning of words. In N. Schmitt &
M. McCarthy (eds.) Vocabulary: Description, Acquisition and Pedagogy. Cambridge: Cambridge University Press, 140–155.
Meara, P. 1996. The dimensions of lexical competence. In G. Brown, K.
Malmkjær & J. Williams (eds.) Performance and Competence in Second Language Acquisition. Cambridge: Cambridge University Press, 35–53.
Mecartty, F. 2000. Lexical and grammatical knowledge in reading and listening comprehension by foreign language learners of Spanish.
Applied Language Learning 11, 323–348.
Namei, S. 2002. The Bilingual Lexicon from a Developmental Perspective.
A Word Association Study of Persian-Swedish Bilinguals. Centre for Research on Bilingualism, Stockholm University.
Namei, S. 2004. Bilingual lexical development: a Persian-Swedish word association study. International Journal of Applied Linguistics 14: 3, 363–388.
Nation, I. S. P. 1990. Teaching and Learning Vocabulary. Massachusetts:
Newbury House.
Nation, I. S. P. 2001. Learning Vocabulary in Another Language.
Cambridge: Cambridge University Press.
Nation, P. and R. Waring 1997. Vocabulary size, text coverage and word lists. In N. Schmitt & M. McCarthy (eds.) Vocabulary: Description, Acquisition and Pedagogy. Cambridge: Cambridge University Press, 6–19.
Newton, J. 1995. Task-based interaction and incidental vocabulary learning: a case study. Second Language Research 11: 2, 159–177.
Nissen, H. B. 2002. Ordassociationer på dansk og engelsk. Unpublished MA-thesis from the Department of English, Germanic and Romance Studies. The University of Copenhagen.
Paribakht, T. S. and M. B. Wesche 1996. Enhancing vocabulary acquisition through reading: a hierarchy of text-related exercise types. The Canadian Modern Language Review 52, 155–178.
Qian, D. D. 1999. Assessing the roles of depth and breadth of vocabulary knowledge in reading comprehension. The Canadian Modern Language Review 56, 282–307.
Qian, D. D. 2002. Investigating the relationship between knowledge and academic reading performance: an assessment perspective. Language Learning 52, 513–536.
Read, J. 1993. The development of a new measure of L2 vocabulary knowledge. Language Testing 10, 355–371.
Read, J. 1998. Validating a test to measure depth of vocabulary
knowledge. In A. Kunnan (ed.)Validation in Language Assessment.
Mahwah, NJ: Lawrence Erlbaum, 41–60.
Read, J. 2000. Assessing Vocabulary. Cambridge: Cambridge University Press.
Read, J. 2004. Plumbing the depths: How should the construct of
vocabulary knowledge be defined? In P. Bogaards & B. Laufer (eds.) Vocabulary in a Second Language. Amsterdam: John Benjamins, 209–227.
Richards, J. 1976. The role of vocabulary teaching. TESOL Quarterly 10, 77–89.
Ringbom, H. 1983. On the distinctions of item learning vs. system learning and receptive competence vs. productive competence in relation to the role of L1 in foreign language learning. In H. Ringbom (ed.) Psycholinguistics and Foreign Language Learning. Publications of the Research Institute of the Åbo Academy Foundation 86, 163– 173.
Schmitt, N. 1998. Quantifying word association responses: what is native- like? System 26, 389–401.
Schoonen, R. & M. Verhallen 1998. Kennis van woorden: de toetsing van diepe woordkennis [Knowledge of words: The testing of deep word knowledge] Pedagogische Studiën 75, 153–168.
Singleton, D. 1999. Exploring the Second Language Mental Lexicon.
Cambridge: Cambridge University Press.
Skriver, S. 2005. Struktur I det mentale leksikon. En undersøgelse af densitet i ordassociationsnetværk. Unpublished MA-thesis from the Department of English, Germanic and Romance Studies. The University of Copenhagen.
Stæhr Jensen, L. 2005. Vocabulary Knowledge and Listening Comprehension in English as aForeign Language: An Empirical Study Employing Data Elicited from Danish EFL Learners.
Unpublished PhD dissertation. The Copenhagen Business School.
Söderman, T. 1993. Word associations of foreign language learners and native speakers: the phenomenon of a shift in response type and its
relevance for lexical development. In H. Ringbom (ed.) Near-Native Proficiency in English. Åbo: Åbo Akademi, 91–182.
Verhallen, M., L. Özdemir, E. Yüksel & R. Schoonen 1999. Woordkennis van Turkse kinderen in de bovenbouw van het basisonderwijs. Een vergelijkend onderzoek bij één- en tweetalige Turkse leerlingen.
Toegepaste Taalwetenschap in Artikelen 61: 1, 21–33. (With an English summary).
Wilks, C. & P. Meara 2002. Untangling word webs: graph theory and the notion of density in second language word association networks.
Second Language Research 18, 303–324.
Wolter, B. 2001. Comparing the L1 and the L2 mental lexicon. Studies in Second Language Acquisition 23, 41–69.
