Abstract

The study tests the retention of two types of words, (a) words that have to be inferred and (b) words glossed in a text in TEXTFUN, an interactive program on the Internet for the practice of reading skills for academic purposes. The results show that retention of the inferred words was much higher than that of the glossed words. This study confirmed theories that the retention of inferred words is higher than the retention of words where the meaning is given, and that incidental vocabulary learning is not particularly efficient. The conclusion of the study is that even an attractive computer program cannot help much in the incidental learning of words.

Introduction

As many teachers of FL reading comprehension will attest, when their students are faced with an unfamiliar text in the foreign language, the first challenge seems to be its vocabulary (Grabe & Stoller, 1997). When the text has many new words, students quickly despair and are discouraged. The connection between vocabulary knowledge and success in reading comprehension tests has been shown in many studies, such as Bossers (1992), Coady (1993), and Grabe and Stoller (1997). When the vocabulary of the text is more familiar, students are more likely to continue with the reading task. [-1-]

Laufer and Sim (1985a, 1985b) show that FL learners seem to rely more on word meaning than on knowledge of the subject or syntax. This means that a certain size of vocabulary has to be known to the learners before they approach a text comfortably. Furthermore, in order to comprehend a text, readers should be familiar with 95 per cent of the words in the text at any level (Hirsch & Nation, 1992). The 'bottom line' for reading English at an academic level is 3,000 words, or 5,000 lexical units, according to Laufer (1992a, 1992b). But in view of the evidence that the vocabulary size of a native speaking university undergraduate is in the range of 14,000-17,000 words (Zechmeister, D'Anna, Hall, Paus & Smith, 1993), Hazenberg and Hulstijn (1996) raise the problem of discrepancy between the minimal size and actual size. Therefore it makes sense to assume that FL teachers should do everything they can, to ensure that their students enlarge the size of their vocabulary.

There are many methods teachers use to teach vocabulary or to encourage vocabulary self-learning by their students. Hulstijn (1992) and Hulstijn, Hollander and Greidanus (1996) distinguish between incidental and intentional vocabulary learning. They claim that both approaches are present in FL learning, since students learn vocabulary intentionally as part of course requirements but also gain knowledge of words incidentally through their reading. [1] Still these authors show that intentional vocabulary learning is more effective for retention. In other words, words learnt intentionally through reading are better retained than words learnt incidentally.

Intentional vocabulary learning, by definition, is intended learning of vocabulary. All other activities that deal with vocabulary are categorized as incidental learning. When students want to increase their vocabulary or have to learn new words for a test, they invest the necessary mental effort and memorize the words until they know their meanings. Incidental learning, on the other hand, does not involve a conscious effort to learn words. This learning "just happens" (but not often). In other words, the number of new words learnt incidentally is relatively small compared to the number of words that can be learned intentionally.

Before continuing, it seems necessary to explain the meaning of "knowing" or "learning" words in a FL. The usual distinction is between active knowledge (knowing how to use the word) and passive knowledge (understanding the meaning of the word). For EFL students, the second, weaker type of knowledge or words is a sufficient learning since they are usually not required to use the words actively in their reading comprehension tasks. More sophisticated distinctions among different types of knowledge of words are made by several authors. One type is presented by Paribakht and Wesche (1997, p. 181), who list five different stages or types of knowledge of words: [-2-]

1. The word is not familiar at all.
2. The word is familiar but its meaning is not known.
3. The meaning is known - the student can supply a correct synonym.
4. The word is used with semantic appropriateness in a sentence.
5. The word is used with semantic appropriateness plus grammatical accuracy in a sentence.

Stage '3' is parallel to "passive knowledge", and as such, is sufficient for EFL students.

Leung and Pikulski (1990) offer another categorization of word knowledge on a 0-3 scale, which they use for testing recognition of words:

0= no recognition of the word meaning.
1= a general, but vague, sense of the word meaning.
2= a good sense of the meaning of the word but not the best meaning.
3= the most appropriate meaning of the word.

Grabe and Stoller (1997, p. 111) suggest another type of categorization of word knowledge in seven stages:

1. Words that are totally unfamiliar.
2. Words where a degree of familiarity can be sensed.
3. Words that must have been in the environment.
4. Words that were familiar and understandable.
5. Words that were familiar and understandable but which presented a need for better, more accurate meanings.
6. Words that had multiple meanings and the most appropriate meaning could be accessed at the time of reading.

Thus it can be seen that 'knowing words' is a subjective concept that depends on the learner's purposes, standards, situation (extensive reading, testing, active usage etc.), teacher or tester's requirements and so on. Still, many researchers claim that incidental vocabulary learning is less efficient in acquiring word knowledge, whatever that is, than intentional learning. Watanabe (1997) claims that "Although incidental learning of vocabulary through context is possible, it is not always efficient" (p. 288). [-3-]

Hulstijn, Hollander and Greidanus (1996) list several reasons for this inefficient learning. These reasons stem from

1. the readers' false belief that they know the words.
2. the readers' decision to ignore the words.
3. the readers' ignorance of the connection between the form of a new word and the meaning contained in the context.
4. the readers' inability to infer a word from context.
5. the non-recurrence of new words (i.e., a single encounter of words).

Not all EFL teachers are aware of the limited efficiency of incidental vocabulary learning. Many of them, especially those who teach at the university, focus on teaching various kinds of reading skills or practicing comprehension questions, and take for granted that their students will somehow learn word meanings. The study emphasizes the importance of intentional vocabulary learning.

In order for students to understand the meaning of a new word, they usually have to either infer it from context or from its structure, or look up the word in the dictionary. Hulstijn (1993) found evidence that the amount of dictionary consultation depends on the task. At the same time, he showed stronger evidence that look-up behavior depends on the relevance of the words. When the subjects in his research deemed words relevant for their reading goal, they looked them up more frequently than words that they considered irrelevant. Hulstijn's subjects were assigned two types of goals in their reading: half of them were required to summarize what they read, and the other half to answer comprehension questions. The results showed that, contrary to expectations, there was no difference in the look-up behavior of the two groups. However, in one paragraph, which was relevant to the comprehension questions' group but irrelevant to the summary group, there was a significant difference in look-up behavior. Subjects in the comprehension questions' group consulted the relevant words much more often than those in the summary group.

The advantage of using the dictionary is that the dictionary use aids retention. Grabe and Stoller (1997) describe a successful case study of L2 learning by a highly motivated learner, who used the bilingual dictionary to study vocabulary both intentionally as well as incidentally. The dictionary not only helped the learner get the accurate meanings of words: "the conscious thought involved in deciding whether or not to look up a word was useful for vocabulary retention" (Grabe & Stoller, 1997, p. 112). [-4-]

Knight (1994) found that students using the dictionary remembered more word meanings than those who had not. In her study, two groups of students, divided according to their level of verbal ability in Spanish (the FL they were studying) were given Spanish texts in the computer and were asked to read them for a recall test. Half of the students in each group had access to an on-line dictionary and half were told to guess meanings from context. All the students were given as much time as needed for their reading. When they finished reading, they got an unexpected vocabulary test on the new words from the texts. It was found that those who used the dictionary spent more time on their reading. Yet, their results were significantly higher than those who had no access to a dictionary. The students with low verbal ability benefited from the dictionary more than those with high verbal ability. Knight's study shows that incidental vocabulary learning occurs. Therefore dictionary use should be encouraged, especially for low-level students, who have poorer inferential skills of guessing from context.

As can be seen in Knight's study, a common alternative to dictionary look-up is inference, which has the advantage that "inferred meanings are remembered slightly better than given meanings" (Hulstijn, Hollander & Greidanus, 1996). Yet the problem with inference is that readers often make erroneous inferences and as a result learn words incorrectly (Bensoussan & Laufer, 1984; Laufer & Sim, 1985a; Hulstijn, 1992, Laufer, 1997). Laufer (1997) explains why these erroneous inferences take place: some words either have a deceptive morphological structure or multiple meanings, 'synforms' (similar lexical forms), to use Laufer's expression. Others are idioms or false cognates. In addition, there are of course words that cannot be inferred or guessed. As a matter of fact, inference is not only an alternative to dictionary look-up: it is also used in the dictionary look-up process, in those cases when the dictionary supplies a list of possibilities with no context or clue what to choose from, and at the same time the context of the word in the text is such that the choice is not easy (Bensoussan, 1983; Koren, 1997). Here, too, the reader has to use his or her inferential skill to decide which meaning is correct, based on their understanding of context.

Another way students can learn vocabulary is by using a glossary, provided one is available to them, of course. This is the easiest way to understand the meanings of words as they appear in context, since it does not even demand the effort of searching and then choosing the appropriate meaning out of several possible ones, which is required by dictionary look-up. But it is doubtful whether using a glossary leads to retention of word meanings in memory. Hulstijn (1992) explains this shortcoming by proposing "a 'mental effort' hypothesis, which predicts that the retention of an inferred word meaning will be higher than the retention of a given word meaning" (p. 113). [-5-]

Another problem with glosses is that they have to be especially prepared by the teacher or writer for each text, or found in specific textbooks, contrary to the dictionary look-up method, which can be done independently by the students. Moreover, a student who constantly depends on a glossary in order to be able to read a text is not likely to become an independent reader. He or she will always need a text especially prepared for them. In this light, a glossary can be a means or a stage in the learning process, but the skilled use of the dictionary as well as good inferring skills should be the next step.

However, hypertext technology enables a quicker and more convenient access to the meanings, as well as other visual and interactive advantages. The availability of hypertext technology enables teachers to teach and test FL texts in a more interesting and efficient manner. For example, Sengupta (1996) describes the creation of hypertext database to teach writing in English, which is simple, attractive and user-friendly. The advantages of hypertext over teaching from a book are as follows:

1. Hypertext enables multiple scannings (quick reading of different texts that are linked together).
2. Users are given many choices (between reading some or all of the linked texts and in the order of the reading the texts autonomously.
3. Users are offered a richly interactive environment (in which they can respond to tasks and questions and get immediate feedback).
4. Hypertext also enables progress at the learner's pace (as each reader works individually without the pressure of finishing classroom tasks together with everybody else).

These advantages cannot be achieved through the linear scanning or use of a textbook which usually demands reading linearly (from beginning to end), does not offer choices as to what or when to read, and, when read in the classroom, requires all students to adapt their speed to the average reader or to the teacher. Aust, Kelley and Roby (1993) describe the use of hyper-reference dictionaries. The advantage of such on-line dictionaries is that they provide immediate access to adjunct information, are quicker than any electronic dictionary, and have a direct return path to the text. Hulstijn (1993) also made use of hypertext technology by using an on-line glossary in order to enable his subjects get meanings of words while reading a text from the screen. His [-6-] subjects got meanings by simply clicking on the words. Due to the current emphasis on rich graphics and animation, all of these hypertext programs (the writing program, the on-line dictionaries and the on-line glossary) are probably both attractive to the user and seem to be rather efficient. This efficiency is however questionable, as will be shown later on.

Not only the learner, but also the researcher, can gain something out of hypertext technology. This technology enabled Hulstijn to see exactly (1) how often the subjects looked up the meanings of the unfamiliar words and (2) which words were looked up. This would have been much more difficult with printed texts since the two possible techniques of "talk aloud" observed by a researcher who takes notes and "videotaping of students" are applicable on a one-to-one basis but are ineffective in big groups of subjects. A computer count of dictionary look-ups is precise and can be used on a great amount of subjects. Another advantage for Hulstijn was that the subjects were not aware of his 'spying' (though they were told about this after the study and accepted the researchers' explanation for this sort of 'unethical' behavior). It must be noted that the 'spying' of the amounts of dictionary look-ups was crucial to Hulstijn's study because this way he could observe correct data during a natural look-up behavior. This could not have been achieved with any other techniques since students' awareness of the researchers' observation or videotaping may have affected their look-up behavior. [2]

I have recently found a way to exploit hypertext technology to the incidental learning of vocabulary (among other skills). Like Hulstijn (1992), who studied the difference between retention of inferred and given (glossed) meanings using conventional printed material, I also studied the difference between inferred and glossed meanings. However, this was done by using the vast possibilities of the computer, which enabled me to overcome some of the drawbacks of the printed method, as will be described below.

The Program

The author recently prepared a hypertext program named TEXTFUN, which can be found on the Internet (http://faculty.biu.ac.il/~korens/textfun). It follows the same lines of user-friendliness, interactivity and interest for the students that were described in the hypertext programs previously mentioned. It is intended for use in a variety of academic levels and for practicing many reading skills in English. The program enables the user to study a text through a variety of reading skills, but to focus on one skill at a time. [-7-]

The user first chooses his or her level, and then a text. Below the name of the text there is a list of reading skills that can be practiced, such as vocabulary in context, syntax, reference, idea-example/support, inference/implications, application etc. The user chooses the skill he or she wants to practice. Then the text appears on the left side of the screen with questions on the right. Parts of the text - words, phrases, sentences or paragraphs - are blue and underlined. At this point, the user has to find the answers to the specific questions using the skill he or she is working on. If the user has difficulty answering the questions, he or she can click on the blue and underlined parts of the text that correspond to the questions to get clues. The user types the answer on the right hand side in the space for 'answer' and can then compare his or her answer with 'correct reply' by clicking on the latter. To reenter the text after clicking for a clue or for a correct reply, the user must click on the "back" button on top of the screen.[3]

The author will only focus here on the vocabulary section of the program. The unfamiliar vocabulary items in the text (those that the author, an experienced EFL teacher of reading comprehension, has assumed that the students are not likely to be familiar with) are divided into two types: inferred and glossed. The author has chosen seven words that can be inferred from context and nine that cannot and are therefore glossed at the end of the text. The synonyms of the glossed words are given in English, and are on a lower level than the glossed words. The words in the text are printed in black, while the words to be inferred are printed in blue and underlined (i.e., linked, as Internet users know well). The glossed words are in red. This way the reader can immediately distinguish between the different types of words. In the vocabulary section there is only one task, requiring the student to infer the meanings of the seven "blue" words and to write them in the white boxes on the right hand side.

If the reader has difficulty inferring the meanings of any of these words, he or she can click it in the text and get a clue (contextual or structural), which is linked through hypertext. After the reader has written the answer in the box on the right, he or she can click on the phrase "correct reply", which is also linked through hypertext, and which appears just below (or next to) the box where he or she wrote the answer. This enables the user to compare his or her answer to the correct answer. [-8-]

As every experienced Internet user knows, the temptation to click on the 'blue' words and see what they are linked to is great, since hypertext links arouse curiosity. The student has the option to avoid using the clues and refrain from comparing his or her answers with "correct replies" given in the program. Yet most students use these hypertext links, even if they do not need the clue and even if they are sure that their reply was correct. This serves the purpose of the program, which is to encourage the students to see the clues and the correct replies, since these two reflect the instructional goals of the program.

By "playing" with the seven words that can be inferred, the students are exposed to the words for a long time and have a higher chance of retention (as was mentioned before). Thus, while each one of the nine glossed words can keep a student busy for about three or four seconds (the time it takes to scroll to the end of the text and find the glossed word and its synonym), a word that needs to be inferred demands his or her attention for at least two or three minutes (the time it takes one to find the clue, think of the possible meaning based on the clue and on the context, write the inferred meaning in the box, find the correct reply and compare it to his or her answer).

The Research

Aims

The research took place after the students used the program for a full teaching session of 90 minutes. The primary aim of the research was to answer the following research questions:

1. Is the retention of inferred words higher than retention of glossed words?
2. How much incidental learning of vocabulary, as is achieved through the program, is retained in the students' memory three days after the learning took place?
3. Is this incidental learning of vocabulary still retained in long term memory (after three months)?
4. Does this program contribute to vocabulary learning? How does it compare to other factors necessary for vocabulary learning? [-9-]

The first research aim was similar to Hulstijn's (1992) objective, which was to compare retention of inferred and given word meanings in incidental learning. The difference between the two research methods is that to overcome the danger of erroneous inferences, Hulstijn used a multiple choice (MC) procedure, which the student had to choose from in his or her attempt to infer. Whereas in the present research, the student was presented with a clue that guided him or her to the correct inference. Both of these techniques limited the student's freedom to infer but guided him towards one correct inference.

The problem with Hulstijn's MC technique, as Hulstijn himself admitted, was that "it was ... found to be inherently error prone: in all experiments there were some subjects for whom some distractors were too appealing" (Hulstijn 1992, p. 123). Hulstijn's solution to this problem was that "the MC procedure should only be used when a teacher is present" (p. 123). This problem was solved in TEXTFUN by 'putting the teacher into the computer', so to speak, i.e. supplying the "correct reply" which prevented wrong inferences and enabled self-practice without requiring the teacher to be present. Although the teacher was present in the computer lesson and helped students with whatever they needed, the help they required was mostly technical (handling the computer, the Internet browser etc.), with few questions about words that were not included in the 16 word list.

Population

The population consisted of 129 students (about 70% females and 30% males) who participated in the computer session. [4] All were EFL university students, most of them first year students, taking an advanced course of four hours a week for a whole academic year. Most of the students were native speakers of Hebrew, with eight native speakers of Russian and two native speakers of Arabic. The average age was 25.

Procedure

Three lessons before the one in which the students were brought to the computer lab, they had been given a pre-test, where they were asked to write the meanings of 16 target words. They were specifically instructed to write every possible meaning they knew for every word. When the pre-test was checked, it was found that they did not know the meaning of any of the words as they appeared in context. Many of them knew one or two words from the list (e.g. 'course' as a noun), but not the same meaning as in the context of the article in TEXTFUN (where, for example, 'course' appeared as a verb). [-10-]

During the lesson on the computer, the students were told to work with as many sections of the TEXTFUN program as they could (they were allowed to choose whatever skills they wanted to practice), but to start with the vocabulary section. They were instructed to read the whole text carefully and to try to understand it because they would need it for the activities of the other sections.

As mentioned previously, the text included 16 target words [5], seven of which (the 'blue' words) were linked to clues and had to be inferred from context and nine (the 'red' words') which were glossed at the end of the text. As described above, the subjects who had to infer the meanings of the 'blue' words could make use of the linked clues and then write the inferred meanings in white boxes on the right hand side of the screen. Once they did that, they could compare what they had written with the correct meaning, which popped up when they clicked the phrase "correct reply" next to the box [6]. The meanings of the glossed words could be found quickly by using the scroll bar downwards. The students were not supposed to do anything with the glossed words except to use them as an aid in their reading. Below is the list of the 16 words, divided according to their function in the program.

[-11-]

Three days after the session in the computer laboratory, during the first lesson the students had after the computer lesson, they were given a vocabulary test. It consisted of one list of the 16 words (the seven inferred and the nine glossed words). They had to recall and write the meaning of each word as it appeared in the context of the article. The task was anonymous, to ensure student cooperation.

The vocabulary test was given in two versions to two groups. The first version consisted of a list of the 16 words and was given to 95 students. The second version consisted of the same 16 words, but each one of them was followed by its original context - the sentence in which it appeared in the text. This test was given to 34 students. The reason the groups were unequal in size is because the original aim was to test retention of words in isolation, since such retention shows a higher level of knowledge of words than when there is dependence on context. When the research was completed, it was decided to compare the results with context-bound words, and see if the poor results found in the words-in-isolation version would also be found in the context version. At that stage, however, fewer students were available.

Results

It was hypothesized that the "blue" (inferred) words would be better remembered than the "red" (glossed) words, and that the students who did the contextual test would succeed more than those who did the non-contextual test. The results of the two tests are presented in Tables 1 and 2 below. In both tests, significant differences between the results of the two types of words were found, with p<0.0001.

[-12-]

Table 1 shows that the students remembered the inferred words much better than they remembered the glossed words. In fact, their retention of the inferred words was four times better than that of the glossed words (33% compared to 8%). This finding corresponds with the theories reviewed before about the conditions for retention. The standard deviations of both types of words are quite similar, which indicates low variance.

Table 2 shows better results than Table 1, which is hardly surprising, considering the nature of the test. Table 2 also shows that the students remembered the inferred words better than the glossed words, but here the gap is about twice as small as the one in Table 1. The meaning of this result is obviously that context helps readers remember the meanings of words, both in learning situations as well as in testing situations. This is true, even when readers have encountered the words only once. The standard deviations of the two types of words are different. The SD of the glossed words is considerably higher than that of the inferred words. This indicates high variance, which means that some students made good use of context to help them recall the meaning of the glossed words and some did not.

Three months after the experiment, the same non-contextual test was given again to 98 students. For the sake of convenience it is called here "the post-test". Here, too, significant differences between the two types of words were found; p<0.008. The results of the post-test are presented in Table 3 below.

N=98	7 inferred words	9 glossed words
Mean	1.2	0.7
%	17	8
S.D.	1.22	1.04

[-13-]

Table 3 shows that after three months the students remembered very little from both types of words. A comparison of the results of Table 3 with those of Table 1 shows that retention of the glossed words was nearly identical: 8% in both tests; whereas retention of the inferred words went down by nearly half (from 33% in the first test down to 17% in the post-test). This can be explained by the assumption that in the course of the three months from the first test to the post-test, the students had had enough time to forget the words. The retention of 8% of the glossed words can be explained as incidental: perhaps they had some opportunities to encounter few of the words in their reading. In any case, 8% is less than one word per student (out of 9), or to put it more accurately, 0.7 word per person.

The gap between the inferred and the glossed words in the three tables confirms the theory that retention of words that the students have dealt with (either by using the dictionary or by putting in mental effort to understand them, as they did in this study with the inferred words), is better than retention of words that they have not dealt with (Hulstijn, 1992; Knight, 1994; Grabe & Stoller, 1997; Koren, 1997). This result has implications for teaching: teachers should prefer to 'activate' their students while they are learning words by giving them controlled inferencing and dictionary assignments. At the same time, the result shows that the whole incidental approach is insufficient. Therefore intentional teaching should be preferred.

It should be noted that the students were not aware of the insufficiency of the program for vocabulary learning. At the end of the lesson with the TEXTFUN program, most students said they had enjoyed the lesson and that it was an interesting and exciting way to learn new words and other reading skills. They also said it had been a refreshing change from the usual classroom sessions, and asked for more lessons of this type. However, their poor results in the first and post-tests show that they did not learn much, in spite of the exciting lesson. [-14-]

Conclusions

The results of this study confirm earlier findings that (1) retention of words learnt through inference from context is better than retention of glossed words, and that (2) incidental vocabulary learning is not very efficient. The focus of the study is on a new program that enables incidental learning in a potentially more interesting way than conventional methods. Yet the findings show that even when new material is attractively presented, and the method enables self-study, interaction, on-line help and encouraging and supporting feedback, retention of new words will not take place without further off-line effort by the learners. In other words, incidental vocabulary learning is not particularly efficient, as shown by the literature. Therefore, intentional learning should rather be encouraged. This view is shared by Paribakht and Wesche (1997), who claim that systematic vocabulary instruction, in addition to learning through reading, is a more successful approach. In view of the pressing need to help the students enhance their vocabulary, it seems that teachers cannot rely on incidental learning and must teach and use strategies that force the students to learn words. This conclusion may have implications for educators and school administrators who invest large sums of money in expensive educational computer programs. If "costly" computer lessons in schools and universities are not followed up with home assignments and class revisions, they may have less value than ordinary classroom lessons which follow these methodological principles.

Notes

References

Aust, R., Kelley, M.J. & Roby, W. (1993). The use of hyper-reference and conventional dictionaries. Educational Technology, Research and Development 41,4, 63-73.

Bensoussan, M. (1983). Dictionaries and tests of EFL reading comprehension. ELT Journal 37,4, 341-345.

Bensoussan, M. & Laufer, B. (1984). Lexical guessing in context in EFL reading comprehension. Journal of Research in Reading 7, 15-32.

Bossers, B. (1992). Reading in Two Languages: A Study of Reading Comprehension in Dutch as a Second Language and in Turkish as a First Language. Rotterdam: Drukkerij Van Driel.

Coady, J. (1993). Research on ESL/EFL vocabulary acquisition: Putting it in context. In: T. Huckin, M. Haynes, & J. Coady (Eds.), Second Language Reading and Vocabulary Learning. Norwood, NJ; Ablex, 3-23. [-16-]

Grabe, W. & Stoller, F.L. (1997). Reading and vocabulary development in a second language: a case study. In: Coady, J. & Huckin, T. (Eds.), Second Language Vocabulary Acquisition. Cambridge: Cambridge University Press, 98-122.

Hazenberg, S. & Hulstijn, J.H. (1996). Defining a minimal receptive second language vocabulary for non-native university students: An empirical investigation. Applied Linguistics 17,2, 145-163.

Hirsch, D. & Nation, P. (1992). What vocabulary size is needed to read unsimplified texts for pleasure? Reading in a Foreign Language 8,2, 689-696.

Hulstijn, J.H (1992). Retention of inferred and given word meanings: Experiments in incidental vocabulary learning. In: P.J.L Arnaud & H. Bejoint (Eds.), Vocabulary and Applied Linguistics, London: Macmillan, 113-125.

Hulstijn, J.H (1993). When do foreign-language readers look up the meaning of unfamiliar words? The influence of task and learner variables. The Modern Language Journal 77,2, 139-147.

Hulstijn, J.H., Hollander, M. & Greidanu, S.T. (1996). Incidental vocabulary learning by advanced foreign language students: The influence of marginal glosses, dictionary use, and reoccurrence of unknown words. The Modern Language Journal 80,3, 327-339.

Knight, S. (1994). Dictionary: the tool of last resort in foreign language reading? A new perspective. Modern Language Journal 78, 285-299.

Koren, S. (1997). Quality versus convenience: Comparison of modern dictionaries from the researcher's, teacher's and learner's points of view. TESL-EJ 2,3, 1-16.

Laufer, B. (1992a). How much lexis is necessary for reading comprehension? In: P. J.L Arnaud & H. Bejoint (Eds.), Vocabulary and Applied Linguistics, London: Macmillan, 126-132. [-17-]

Laufer, B. (1992b). Reading in a foreign language? How does L2 lexical knowledge interact with the reader's general academic ability? Journal of Research in Reading 15,2, 95-103.

Laufer, B. (1997). The lexical plight in second language reading: words you don't know, words you think you know, and words you can't guess. In: Huckin, T. (Eds.), Second Language Vocabulary Acquisition. Cambridge: Cambridge University Press, 20-34.

Laufer, B. & Sim, D.D. (1985a). Taking the easy way out: Non-use and misuse of clues in EFL reading. English Teaching Forum 23,2, 7-10.

Laufer, B. & Sim, D.D. (1985b). Measuring and explaining the threshold needed for English for Academic Purposes texts. Foreign Language Annals 1,8, 405-413.

Leung, C.B. & Pikulski, J.J. (1990). Incidental learning of word meanings by kindergarten and first-grade children through repeated read aloud events. In: J. Zutell & S. McCormick (Eds.), Literacy Theory and Research: Analyses from Multiple Paradigms. Chicago, Il: National Reading Conference. 231-239. From: Grabe, W. & Stoller, F.L., p. 105.

Meara, P. & Buxton, B. (1987). An alternative to multiple choice vocabulary tests. Language Testing 4, 142-154.

Paribakht, T.S. & Wesche, M. (1997). Vocabulary enhancement activities and reading for meaning in second language vocabulary acquisition. In Coady J. & Huckin, T. (Eds.), 174-200.

Sengupta, S. (1996). Creating a hypertext database to help Hong Kong English language teachers teach writing. System 24, 187-198.

Watanabe, Y. (1997). Input, intake and retention: Effects of increased learning of foreign language vocabulary. Studies in Second Language Acquisition 19, 287-301.

Zechmeister, E.B., D'Anna, C., Hall, J.W., Paus, C.H. & Smith, J.A. (1993). Metacognitive and other knowledge about the mental lexicon: Do we know how many words we know? Applied Linguistics 14,2, 188-206.

About the Author

Dr. Shira Koren is a member of the senior staff at the EFL (English as a Foreign Language) Unit, Bar Ilan University, Ramat-Gan, Israel. Her research interests include pronunciation acquisition, musicality and pronunciation, attitude to FL, motivation, dictionaries, vocabulary acquisition, note-taking in FL and children's literature (fairy tales). Dr. Koren is also an author of children's books.