Acquiring vocabulary through reading: Effects of frequency and contextual richness

 

 

Rick Zahar (Langara College, Vancouver)
Tom Cobb (Université du Québec à Montréal),
Nina Spada (OISE, University of Toronto)

Published in the Canadian Modern Language Review - Volume 57, No. 4, June / juin 2001

 

Web Pre-publication

This paper is based on the MA project of the first author. Address correspondence to the second author at cobb.tom@uqam.ca .

 

Abstract: While L2 vocabulary acquisition research is no longer "a neglected area" (Meara, 1980), a lack of  progress remains on some basic questions. One concerns the number of times a word must be encountered to be learned. Even using similar learning criteria, estimates range from six (Saragi, Nation & Meister, 1978) to 20 (Herman, Anderson, Pearson, & Nagy, 1987). Another question concerns the types of contexts that are conducive to learning. Some studies have reported that rich, informative contexts are the most conducive to acquisition (Schouten-van Parreren, 1989); others that rich contexts divert attention from the lexical level and produce little acquisition (Mondria & Wit-deBoer, 1991). These phenomena were investigated in a vocabulary acquisition study with Quebec school-aged ESL learners at five levels of proficiency. First, learners read a text and were tested on its new vocabulary.  Then, learned and unlearned words were compared for frequency of occurrence and level of contextual support. Frequency needs were found to be related to learner level, and contextual richness was unrelated to learning.

Résumé : Bien que la recherche sur l’acquisition de vocabulaire en L2 n’est plus un «domaine négligé» (Meara, 1980), son progrès relativement à des éléments de base reste peu avancé. Par exemple, nous ne savons toujours pas le nombre de fois qu’un apprenant doit rencontrer un mot pour l’apprendre. Même en utilisant des critères d’apprentissage semblables, les estimations varient entre six (Saragi, Nation et Meister, 1978) et 20 fois (Herman, Anderson, Pearson et Nagy, 1987). Il s’agit également du type de contexte qui se prêtent à l’apprentissage. Plusieurs études soutiennent que des contextes riches et informatifs sont les plus propices à l’acquisition de vocabulaire (Schouten-van Parreren, 1989), alors que d’autres affirment que les contextes trop riches détournent l’attention du niveau lexical et génèrent peu d’acquisition (Mondria et Wit-deBoer, 1991). Ces phénomènes ont fait l’objet d’une étude de recherche sur l’acquisition effectuée auprès d’apprenants scolaires au Québec à cinq niveaux de compétence différents. Les apprenants devaient lire d’abord un texte, et ensuite passer un test sur le nouveau vocabulaire. Les mots qu’ils ont appris étaient comparés avec les mots qu’ils n’ont pas appris quant à leur fréquence d’occurrence et leur niveau d’appui contextuel. Les résultats ont démontré que les exigences en fréquence avaient un rapport avec le niveau de l’apprenant, et que la richesse contextuelle n’avait aucun rapport avec l’apprentissage.

 

Interest in second language vocabulary acquisition (SLVA) has grown steadily in the last ten years. It has produced several excellent books (Huckin, Haynes, & Coady, 1993; Huckin & Coady, 1997; Nation, 1990; Schmitt & McCarthy, 1997; Singleton, 1999; Schmitt, 2000), special issues of journals (including The Canadian Modern Language Review, Harley, 1996; Second Language Research, Meara, 1995; and Studies in Second Language Acquisition, Wesche & Paribakht, 1999), and a growing number of research articles investigating ever finer details of word learning, storing, and processing (many of which have been assembled by Paul Meara on  the VARGA website). Until recently SLVA was a special interest group whose members knew each other and who had read most of what there was to read on the subject; now it is another large subdivision of applied linguistics producing more books and papers than anyone can keep up with. And yet, in the midst of a growing satisfaction with the field’s maturity and acceptance, a number of rather basic questions about SLVA remain unanswered, and the impact of vocabulary research on vocabulary pedagogy has been rather limited (Singleton, 1997), possibly because these basic questions remain unanswered.

One important question which remains unresolved concerns whether a functional reading lexicon in a second language (L2) can itself be acquired through reading (Nagy, 1997), or is more likely to result from some kind of direct instruction (Zimmerman, 1997) or instructionally enhanced reading (Hulstijn, Hollander & Greidanus, 1996). One part of the answer to this question depends on what is meant by a functional reading lexicon, i.e. the minimal recognition vocabulary knowledge needed to facilitate reading comprehension. One often cited study defines this lexicon as 3000 word families chosen on a basis of simple frequency (Laufer, 1988) (Note 1). The other part of the answer depends on how many words learners can be expected to acquire in a given period of time and whether there is any reliability to this learning rate. A British study by Milton and Meara (1995) found that learners studying in the foreign language environment could learn at a rate of 2500 words per year, while the rate of those studying in classrooms at home in the L1 environment was only about 550. However, these are the only longitudinal figures available on this question and the finding has never been tested through replication. The more typical investigation of learning rate involves the pretest-posttest reading study, where learners read a short text and then are tested for the number of new words they acquired from it. On this measure there appears to be some degree of replicability, as can be seen in the review of several such studies in Horst, Cobb, and Meara (1998). After reading, participants can typically select a definition for a little more than one out of every 12 words tested. Does this figure suggest that a functional lexicon can be acquired through reading, or not? One purpose of the present study is to subject the one-in-12 figure to the test of replication, and to extrapolate this rate in consideration of the size of text and the time typically available for second language learning. 

Logically prior to the question as to whether vocabulary acquisition is most effective via direct versus indirect instruction/exposure are a number of smaller unanswered questions about the conditions under which lexical acquisition from normal reading takes place. One is a question about word frequency, the number of times a word must be encountered in reading for it to be learned. If we knew that words were learned in 50 encounters but that frequent words might appear only twice in a general corpus of a million words, then we might suspect that a reading lexicon would not be acquired through reading. What does the research say about the number of encounters needed to learn a word? Saragi, Nation and Meister (1978) found that words presented to learners fewer than six times were learned by half their subjects, while words presented six or more times were learned by 93%, suggesting a threshold of six encounters. In a review paper a few years later, Nation (1982) found that a figure of 16 encounters was common in the literature. In an empirical study, Jenkins, Stein and Wysocki (1984) found that only about 25% of learners had learned a word after 10 encounters. Working with a different metric, Nagy, Herman and Anderson (1985) estimated the likelihood of acquisition after one encounter to be about .15 (with full acquisition occurring in six or seven encounters?) In a follow-up to this experiment, however, Herman, Anderson, Pearson, and Nagy (1987) revised the estimate downward to .05 for authentic texts (full acquisition in 20 encounters?) A more recent study by Rott (1999) using experimental texts located the critical watershed at six encounters, bringing us all the way round to Saragi et al's (1978) figure. The list of conflicting measures and measurements could be extended almost indefinitely; it seems safe to say that there have been no definitive studies which show how many exposures to a word are needed to promote its learning. Without an answer to this rather basic question, the direct vs. indirect vocabulary learning debate is unlikely to advance.

Part of the problem with measuring number of encounters is that word learning is incremental, so that a single encounter with a word may provide some amount of learning, while 100 encounters will still not engender a native speaker’s complex knowledge of the word (its collocations, associations, and pragmatic values). In other words, researchers may be defining and measuring ‘learning’ in different ways. However, even if the test of learning is reduced to ‘ability to select a short definition for the word on a multiple choice test,’ which is the format of the most widely used standard test (Nation's, 1990, Vocabulary Levels Test) and of the majority of non-standard tests reported in the vocabulary research literature (e.g., the studies reviewed in Horst, Cobb & Meara, 1998), there is still little agreement about how many meetings are needed on average to produce this knowledge.

A possible clue to the source of the variance in frequency estimates may be that different numbers of encounters are needed by different levels of learners. In two separate studies, Horst (1998, 2000) investigated frequency of encounter and level of proficiency with two groups of English as a foreign language (EFL) learners in two countries. Both studies involved determining which words the students would learn from reading a text and how this related to the number of times the words had occurred in the text. The studies presented contradictory findings. For one group of learners, there was a substantial correlation (.49) between the words learned and the number of occurrences (reported in Horst et al, 1998), but for a second groups of learners there was no significant correlation (Horst, 2000). A closer examination of this disparity revealed that the two groups were not comparable in terms of their overall vocabulary size. The group who knew more words (as established by Nation’s, 1990, Levels Test) needed fewer encounters to learn another word, while learners who knew fewer words required more encounters. It makes sense that learners with more words should need fewer encounters to learn another word, if only because they will know a higher proportion of the words in the vicinity of the new word, and conversely that learners with fewer words will wait longer for a context that is clear to them. A second purpose of the present study is to clarify the frequency question through a replication of Horst’s research within a single learner context and population.

            The kinds of contexts that facilitate the acquisition of new words is another issue that remains unresolved. For example, are words more successfully learned from clear, from semi-clear, or from unclear contexts? The background for this question is Beck, McKeown and McCaslin's (1983) finding that natural texts contain a high proportion of contexts that are unsupportive or even misleading for word learning purposes. Most teachers and many researchers (e.g., Schouten-van Parreren, 1989; Beheydt, 1990) regard as obvious that clear, rich, and supportive contexts are conducive to learning, even to the point of recommending that teachers design texts with especially clear contexts specifically for word learning. In other words, learners should not be allowed to flounder in the variable contexts provided by natural text, or at least should not depend on such floundering for their vocabulary growth. However, there are research findings that challenge this assumption. Studies by Mondria and Wit-deBoer (1991) and by Parry (1991) found that supportive contexts facilitated reading more than they facilitated word learning. It appears that learners do not necessarily notice or remember new words if these are fully redundant with their contexts. In other words, slightly less helpful contexts, or a mix of context types, could slow down reading but speed up vocabulary acquisition, and normal reading would be an ideal way to meet such a range of contexts. A third purpose of this study is to examine the types of contexts that learners do and do not learn words from when reading.

            To summarize, while a great deal of progress has been made in SLVA research, basic questions related to the acquisition of vocabulary from text remain unanswered. It is important that such questions should be answered, since it is precisely in text that the great majority of English words have their usual and often only home. Three of the most fundamental of these questions concern whether building a reading lexicon through reading is even feasible in principle, how many occurrences are needed to learn a new word, and what type of contextual support is needed to learn a new word. The research reported here attempts to make some progress on these related questions by examining them within the context of an empirical study investigating the contributions of reading to incidental vocabulary acquisition. Thus, the research questions in this study are:

 

(1) Is the amount of L2 vocabulary acquisition that takes place through reading (a) predictable, and (b) sufficient in principle to establish a functional lexicon           

 

(2) Does the number of encounters needed to learn a word vary with learner proficiency?

 

(3)  Does type of contextual richness contribute differently to vocabulary acquisition?

 

To investigate these questions, the study followed a pretest-posttest experimental model which included (1) testing the participants’ vocabulary size, (2) selecting an appropriate text based on the participants’ vocabulary level, (3) choosing words from this text that the students would be unlikely to know, (4) testing knowledge of these words in a pretest (and revising the selection where necessary), (5) having participants read the text under controlled conditions, (6) giving the pretest again as a post-test, and (7) calculating and extrapolating the amount of vocabulary acquisition gained from reading, and (8) carrying out a post-hoc analysis to identify the conditions under which acquisition occurred, with respect to the number of occurrences and the type of contextual support of the particular words learned.

 

Context of the study

 

The experiment was carried out in the classrooms of a private Montreal French-language high school for boys. In this school, students receive a total of 75 hours of ESL instruction during the academic year. (Note 2) The type of instruction provided is based on a communicative approach in which the emphasis is on meaning rather than form. Consistent with the philosophy and guidelines of the Quebec Ministry of Education (MEQ), there is equal emphasis placed on oral and written production as well as listening and reading comprehension. There is no systematic vocabulary instruction throughout the levels and grades. Each spring, the newly accepted students for the following year participate in an English placement exam to ascertain their English language competence level. The test is designed to place them in one of five streams ranging from beginner to bilingual. The test has been developed over the years by the school’s English department and it is a combination of production and comprehension tasks. It does not test vocabulary separately.

 

Participants

 

The participants were 144 male grade seven ESL students. They had previously been placed into five streams of ESL study ranging from beginner to bilingual. They had completed three semesters of class time prior to the commencement of this experiment (one and a half years, with 75 hours of English instruction per year). The students came from a variety of first language backgrounds but were mainly Francophones. The school’s general academic instruction was in French, and all students had received at least some ESL instruction prior to entering the school. Because the students came from many different schools prior to entering this school and grade, the extent and nature of their previous ESL instruction is unknown.

 

Procedures and Results

 

The Levels test

 

Pencil and paper versions of Nation’s (1990) Vocabulary Levels Test were given to the students, with 45 minutes to complete the test. This simple test asked students to match words to brief definitions at five frequency levels (see Appendix 1 for a sample of the test, and Note 3 for some indications about the test's limitations). Scores were recorded for each student for each of the five levels of Nation's test, and group mean score were subjected to an analysis of variance to determine the significance of any differences between classes. The results of these analyses are presented in Table 1.

 

TABLE 1

Percentage Scores on Levels Test by Ability

Group

Vocabulary Level

 

2000

3000

5000

UWL

10000

Total

1 (low)

50

56

39

33

17

38%

2

61

72

44

39

22

48%

3

72

83

56

56

39

60%

4

83

89

67

62

39

69%

5 (high)

94

100

83

72

56

80%

Note. UWL = University Word List

 

Looking at Table 1, we can see that a clear pattern emerges. Students in all groups are most proficient at the 2000 and 3000 word levels, with Group 1 knowing approximately one-half of the words at these levels and Group 5 knowing most of them. It is interesting that scores are higher at the lower frequency 3000 level than at the 2000 level; what this means is not clear, except that these learners are probably not moving through a structured program of vocabulary growth. None the less, the mean for each group increases linearly with group placement level, with overall differences statistically significant (F(4, 153)= 89.09, p< .001), (Note 4) as well as all pairwise differences (following a procedure for post-hoc means comparison discussed in Woods, Fletcher, & Hughes, 1986, p. 210) with the exception of groups 4 and 5. Thus the school’s placement test distinguishes the students in terms of initial vocabulary knowledge.

 

Story Selection

 

The task of finding a story that was of a vocabulary level low enough that the beginner group could understand, but also challenging enough to have some vocabulary that the higher proficiency groups would not know was not easy because the vocabulary scores for Group 1 started at 50% of the 2000-word list and rose to 95% for Group 5. A story entitled The Golden Fleece (see Appendix 2) was chosen from an intermediate ESL reader used by Groups 3 and 4 of this grade at this school. The story comes from an authentic textbook that has been graded for students of this age and proficiency level (although it had not been used by any of these students previously), and it is a Greek myth and thus contains some uncommon vocabulary.

In order to verify the suitability of the vocabulary of the text, the story was scanned by computer and analyzed with concordance (Zimmerman, 1988) and lexical profiling (Cobb, 1998) software. The concordance located all similar words in the text and displayed them in their contexts, and the lexical profiler broke the text down into Nation’s word frequency categories. These categories are identical to the main levels of Nation’s test, and break down as follows: 1) function words such as the, and, of; 2) first 1000 words, such as days, deep, fear and field; 3) second 1000 words like applaud, beast, breath and fierce; 4) University Word List terms such as hero, myth, tasks and withdraw; and 5) off-list words, i.e., less common words that are not on any of these list. The purpose of this profiling analysis was to assess whether or not The Golden Fleece fell within the vocabulary range of the participants. (The purpose of the concordance analysis is discussed below.)

 

TABLE 2

Lexical composition of the stimulus text

Function words

54%

First 1000 words

29%

Second 1000 words

08%

UWL

01%

Off-list words

08%

 

Table 2 shows the lexical profile of the stimulus text and suggests it was suitable for the present experiment in three important ways. First, 91 per cent of the words were from the 0-2000 word frequency range, making the text relatively unsophisticated and thus appropriate for the majority of students in the lower groups. Second, the text nonetheless provided a relatively high 8 per cent of off-list words that even the more advanced students would be unlikely to know. Third, the text contained a very low proportion of University Word List items, which are often cognate with French and can be inferred on a basis of simple similarity.

Thirty words from The Golden Fleece of various frequency levels and numbers of occurrences in the text were chosen for the pretest (see Appendix 3). Of the 30 words, 37% came from the 0-1000 word list, 30% from the 1000-2000 list, and 33% were off-list or low frequency. The pretest design was similar to the Levels Test so that subjects were already familiar with the test task, which was to match words to brief definitions. The entire test was run through the lexical profile program, and all items examined to ensure that the definitions did not contain less common words than the words they were defining. An effort was made to choose words that were not cognate with French (with one exception, oracle) and hence inferable on a basis of similarity.

 

Pretest Results

 

Pretest scores were analyzed to determine the mean score of each class for the entire test, and to test the significance of differences between groups. This was done to determine whether, as with the initial Levels Test, the students’ knowledge of the test items would correlate with their placement group. The results are presented in Table 3.

 

TABLE 3
Pretest Means  (N=144)

Group

N

M/30

SD

1

2

3

4

5

22

27

34

32

29

10.77

16.56

20.03

24.40

26.55

.71

.64

.57

.59

.62

 

The students’ mean test averages increased as group level increased and there was a significant main effect for level (F(4, 139) = 90.32, p< .001), with all pairwise distinctions once again significant with the exception of the difference between groups 4 and 5. In other words, pre-story specific vocabulary knowledge parallels the vocabulary knowledge of English as a whole as assessed by the Levels Test.

 

Treatment Procedure

 

Thirteen days after the pretest, the treatment began. During the 13-day interval, the students were on vacation, and there is little likelihood they would have had further exposure to any of the test words. The delay was also intended to allow a lapse between seeing the words in the test and seeing them again in the story, reducing any possibility that test words would have been learned from the pretest definitions rather than from the story itself.

In every class, the students initially heard the story on cassette tape while they followed along simultaneously with the written text. The reasoning for this procedure was based on the observation by Horst, Cobb and Meara (1998), that hearing as well as reading the text would ensure that there was at least one complete input of the story for the lower level readers. It was also assumed that the circumstances of following along as a text is read aloud would preclude opportunities for intentional word learning, looking words up in a dictionary, jotting down words for look-up later, and so on. Shefelbine (1990) further adds that this presentation mode is a way of eliminating the confounding effects of poor decoding ability. Subjects were not allowed to ask for any clarification of words from the tester throughout the reading as this was a test of acquisition from reading and questions might have influenced their vocabulary comprehension over and above what they might have derived from the text itself. They were simply told to try their best to understand as much as possible. After the story was heard on tape, subjects were given the remainder of the period to reread the story as many times as they wanted to or felt were necessary. The classroom teachers who participated in the experiment reported that students said they had enough time to re-read the story once or twice more on their own.

 

            Posttest Results

 

The posttest was given in the next class meeting two days later. The post-test was the same as the pretest and given under the same conditions. The hypothesis was that there would be a significant gain in vocabulary from pretest to posttest across groups. The posttest results were analyzed in the same manner as the pretests and are presented in Table 4.

 

TABLE 4

Pretest to Posttest Gains by Group (N=144)

Group

N

Pretest

(SD)

%

Post

(SD)

(%)

Gain

(%)

1

22

10.77

(.71)

36

12.59  

(.74)

42

1.82

06

2

27

16.55

(.64)

55

19.59

(.66)

65

3.03

10

3

34

20.02

(.57)

67

22.38

(.59)

74

2.36

07

4

32

24.41

(.59)

81

26.53

(.61)

88

2.12

07

5

29

26.55

(.62)

89

28.03

(.64)

93

1.48

04

 

When we compare the mean scores from pretest to posttest in Table 4, we can see a positive change in each group’s means. The average number of words learned across groups was 2.16 from the 30 words tested. Group 5 learned least in absolute terms (with an average of 1.48 words), while Group 2 learned most (with an average of 3.03 words). Analysis of variance revealed a significant main effect for time between pretest and post-test (F(4,139)=96.88, p< 0.001), with pairwise differences once again significant apart from the difference between Groups 4 and 5.

A comment on Groups 5's performance may be in order. Although these learners may have learned least in absolute terms, this is merely due to a ceiling effect in the present test. Group 5 participants learned most in terms of proportion of learning opportunity. There were only 3.45 words for these students to learn, of which they learned 1.48, or 43%. In contrast, Group 2 learned 3.03 words from a learning opportunity of 13.45, or 22.5%, about half. The present study, however, was not prepared to investigate relative learning phenomena (for a discussion of these see Shefelbine, 1990). For lack of variance in absolute gains, and the repeated lack of a statistical distinction between Groups 4 and 5, it was decided to eliminate Group 5 from the remaining analysis.

To summarize, these results support the findings of previous research showing modest but apparently replicable learning gains. In the studies reviewed by Horst, Cobb and Meara (1998), the average learning gain was one word for every 12 tested. In this study, with Group 5 eliminated the average learning gain was 2.33 words for 30 tested, or .93 for 12. The next phase of this research involved a post-hoc investigation of the extent to which frequency and context type had contributed to acquisition.

 

 

            The Role of Frequency

 

Each student’s pretest and posttest results were compared to identify all the words that had been acquired as a result of one hour with The Golden Fleece. The words on the test, their frequency in the text, and the number of students from each group that learned them are presented in Table 5.

TABLE 5
Test Words by Frequency and by Groups Acquiring Them

Test

Words 

Frequency

in Text

 

Acquired Words by Group

Total subjects learning each word

 

 

1

2

3

4

5

 

 

 

 

 

 

 

 

 

angrily

3

5

5

2

1

1

14

breath

1

5

6

4

2

0

17

centaur

7

5

5

5

5

4

24

crash

3

3

6

2

3

1

15

crew

3

4

3

4

3

0

14

daughter

4

1

1

0

0

0

2

dove

5

3

7

10

6

2

28

drew

4

1

2

6

4

3

16

fearful

4

4

5

4

0

1

14

field

7

4

3

2

0

0

9

fight

2

4

0

1

0

0

5

fleece

15

5

7

6

5

3

26

forth

2

1

4

10

7

1

23

helmet

3

4

1

1

0

0

6

manhood

2

3

8

2

1

0

14

oars

2

1

2

2

4

2

11

oracle

9

6

9

10

8

9

42

plow

4

6

6

10

7

4

33

rightful

4

5

7

1

2

1

16

ruler

5

2

4

4

0

0

10

rush

8

4

2

2

1

0

9

sheep

2

3

3

3

0

3

12

sow

2

2

3

6

9

6

26

spears

2

4

4

1

2

0

11

stream

2

2

5

5

2

0

14

swift

2

4

10

10

2

1

27

sword

2

1

4

1

0

0

6

teeth

4

4

7

2

4

2

19

wisdom

2

1

3

4

1

0

9

youth

3

4

3

2

2

3

14

 

 

Data on the number of times each word appeared in the text and the number of students in each group who had learned the word were entered into a multiple regression analysis, producing a correlation between frequency and learning gain for each of the four remaining groups. It will be remembered that the groups’ vocabulary size increases linearly.

 

TABLE 6

Correlations Between Frequency and Learning Across Groups

Group

Freq x Learning (r)

Variance accounted for by frequency (R2)

1-4

.36

12.9 %

1

.40

16.0 %

2

.22

 4.8 %

3

.21

 4.4 %

4

.24

 5.7 %

 

Overall, frequency and learning correlate reasonably well at r=.36, with frequency accounting for 13% of learning variance. However, most of this effect pertains to Group 1, the learners with the smallest vocabulary sizes. Frequency plays a greater role in acquisition at the lowest level (r=.40, R2= 16%), and thereafter plays a consistently lesser role (accounting for R2 = 4.8%, 4.4%, and 5.7% of variance respectively in the higher groups). In other words, frequency appears to be three to four times more important for beginners as it is for more advanced students. This suggests that learners who know fewer words need to meet a new word several times before they have learned it, even to a passive level as measured in the present posttest. On the other hand, learners who know more words seem able to accomplish the same amount of learning in fewer occurrences.

 

            The Role of Contextual Richness

 

To investigate the degree of contextual richness that led to greatest learning, we limited the analysis to the most and least acquired words from the 30 targets, which had been (1) consistently learned or not learned across groups, and (2) for which there was full learning opportunity for the highest group, Group 4 (i.e. they were not known at pretest). Under these provisos, the 30 test words were reduced to a subset of 10 words, six most acquired words, and four least acquired. Most acquired were centaur (n.), dove (n.), fleece (n), oracle (n), plow (n, v), and sow (v); least acquired were drew (v, i.e. pulled), oars (n), sheep (n), and youth (n).

These ten words were analyzed for their contextual richness and number of appearances using Beck et al’s (1983) model of contextual support level aided by a computer concordance of The Golden Fleece. The concordance listed each appearance of a target word along with its immediate context, and from this presentation two of the researchers assessed the contextual richness of each word and categorized it according to Beck et al’s (1983) context levels (see Table 7). These levels constitute a four-point nominal scale, which is as follows:

 

1) Misdirective contexts (which seem to indicate an incorrect meaning. Example: “He has a frog in his throat.” (None of these occurred in The Golden Fleece text.)

 

 2) Nondirective contexts (which do not seem to direct the reader toward any particular meaning). Example (from the text): “Then they released the dove.”

 

 3) General contexts (which provide enough information to place the word in a general category). Example (from the text): “But the dove can fly through swiftly enough to escape destruction.”

 

4) Directive contexts; which present a word with an implicit definition so that it is understood. Example (from the text): “A centaur was a strange creature, half man and half horse.”

 

By way of illustration, some of the judgments of the contextual support levels (CSL's) supporting centaur appear below. Each context was judged by two raters (with interrater reliability = .95).


1) A centaur was a strange creature, half man and half horse. CSL 4

 

2) From the waist up, the centaur was a man, gifted with intelligence. CSL 3

 

3) Below the waist, the centaur had the four legs and body of a horse, which gave

him the speed and strength of that animal. CSL 3

 

4) But when the young prince reached manhood, Chiron the centaur was satisfied that Jason was fit to be a king. CSL 2

 

5) He grew from babyhood into boyhood; and when it came time for Jason to go to school, he was sent to study with the centaurs. CSL 2

 

6) The centaurs were famous for being very learned. CSL 2

 

7) Young Jason's special teacher was Chiron, the chief of the centaurs. CSL 2

 

The results of the contextual support level analysis of the most and least acquired words are presented in Table 7, showing frequency plus means and standard deviations of contextual richness.

 

TABLE 7

The Most and Least Acquired Words, their Frequency and Mean Context Ratings

 Most acquired

Frequency

Richness Ratings

Mean

s.d.

centaur (n)

7

4,3,3,2,2,2,2     

2.60    

.58

dove (n)

5

3,3,3,2,2

2.60   

.48

fleece (n)

15

4,4,2,2,2,2,2,2, 2,2,2,2,2,2,2

2.30    

.56

oracle (n)

9

4,3,3,3,2,2,2,2,2

2.50    

.86

plow (n)(v)

4

3,3,3,2          

2.75    

.44

sow (v)

2

3,2           

2.50    

.50

 

 

 

 

 

 Least acquired

Frequency

Richness Ratings

Mean

s.d.

drew (v)

4

4,3,3,3         

3.25 

.44

oars (n)

2

2,2           

2.00      

.00

sheep (n)

2

2,2           

2.00    

.00

youth (n)

3

4,3,3          

3.30    

.44

 

 

            Results of the context analysis

 

The first pattern to note in Table 7 is that, in general and across groups, learned words tend to be those that appear most frequently. The average frequency of most learned words is seven (s.d.= 4.6), and of the least learned words it is 2.75 (s.d= .96). The second observation to be made is that given enough frequency, the contextual support levels tend to boil down to a consistent average of about 2.5, with a standard deviation about .5. In other words, if this text is typical, then the problem of what sorts of contexts do and do not support vocabulary learning could be a non-issue. If students read texts that are roughly tuned to their level, then a suitable range of context types will appear for each word as a statistical feature of natural text, provided, of course, that the students meet the word enough times. By this account, then, the effect of contextual support appears to be subordinate to frequency.

This result requires replication with other texts and learners before it is fully accepted. For one thing, there are no ‘misdirective’ contexts in the particular story used in this experiment, and although the story is designed for ESL students one could not take for granted that such texts would never contain this feature. Indeed, Beck et al (1983) found many misdirective contexts in their L1 subjects’ materials, so the finding may apply mainly to ESL materials (however, see additional data below). For another thing, frequency and contextual richness were examined in a post hoc analysis in this study, and further studies are now needed to investigate more systematically the specific interactions between these variables. The findings in this study suggest some promising directions for research in this domain.

 

Discussion and Conclusion

 

The purpose of this study was to investigate three interrelated questions about vocabulary growth through reading: Is there a replicable learning rate, and if so does it support the acquisition of a reading lexicon through reading? Is the number of occurrences needed to learn a word a function of learners' lexical level? And is there a contextual support level that seems most productive of acquisition?

 

Contextual richness

 

Starting with the last question first, our data suggests that there is no difference between the contextual informativeness of words learned by all participants and words learned by none. Words appear in natural text in a uniform variety of context types, so that with a modicum of frequency (and in interaction with the learner's prior knowledge) all of these context types will be presented to the learner and the word is likely to be learned. The role of this variability is not clear. Perhaps it ensures that the learner finally meets the word in a one or two clear contexts that it can be learned from, or perhaps it provides the mix of clear and unclear contexts that a word can be learned from. Given that learners apparently do not learn very well from rich contexts (Mondria et al, 1990), and can not learn anything at all from totally opaque contexts, it could be in the alternation of more and less informative contexts, as appears to characterize natural text, that words can be learned. A speculative mechanism might be as follows: an unclear or semi-clear context opens up a learning need, or conceptual gap (Brown, 1993), which is then reactivated when the word is eventually met in a clear context.

             To check whether uniform distribution of context types might be mainly a feature of short texts designed for language learners, such as the Golden Fleece text used here, Horst (2000) replicated this part of our experiment with longer texts written for native speakers. In two separate studies, she had subjects read authentic texts of 6,000 and 9,500 words, with learning tasks of 114 and 136 previously unknown words, respectively. In both cases the target words were words which appeared only once, where context effects might have been predicted. Using the same computer aided version of Beck, McKeown and McCaslin's (1983) four-way classification, Horst recorded the context type for each to-be-learned word and compared it to whether the word had in fact been learned. Like us, she found no relationship between words learned and the type of contexts they appeared in. In her first experiment, contextual support for learned words was 2.6 (s.d. =60) and for unlearned words 2.68 (s.d.=66). Similar results obtained for the second experiment.

 

Frequency of occurrence

On the matter of the inconclusive frequency findings typifying the research literature to date, our study successfully replicates those of Horst (Horst et al, 1998; Horst , 2000) but within a single learner population. It seems clear that weaker learners are indeed more dependent on frequency than stronger ones, and that learner level should be taken into account in the design of future frequency studies. Our data adds some detail to Horst's finding, showing frequency to be more than three times as determining for learners scoring 50% on the 2000 level of the Levels Test than for learners scoring 60%, at least in the case of a text with most (91%) of its lexis drawn from the that level. In other words, there is a hint of some sort of lexical threshold or sub-threshold just after half way through the 2000 frequency zone, which is coincidentally about where lexis that is more typical of spoken English gives way to lexis which is more typical of written English. This finding should be replicated with other learners and other types of texts, but for the moment it seems clear that looking at frequency in terms of learner level will offer a useful approach to solving the frequency mystery. The mystery could possibly be solved in the next few years through the inclusion of a standard measure of prior lexical knowledge as a covariate in frequency studies (although the fact that we do not as yet have adequate standard measures in the SLVA field, as lamented by Meara, e.g., 1996, could limit the success of such a project.) If we knew whether there were stable frequency requirements of different levels of learner and what these levels were, this would help to eliminate some of the guesswork in the design of instructional materials for language learning.

 

Direct vs. indirect instruction

In the introduction it was proposed that questions about frequency and context type were logically prior to the larger question of direct versus indirect instruction. How do our findings bear upon this larger question? In one sense, our findings are favorable to the idea of building a reading lexicon through reading. Natural text seems to be remarkably well designed for learning purposes, which is not surprising since most native speakers of a language are able to expand their vocabularies through reading. As already discussed, variation in contextual support level appears to be remarkably uniform, and we have speculated that this variation may have a good deal to do with vocabulary acquisition. Somewhat less speculative is the way frequency in learning and frequency in a natural language could fit together. Beginning learners apparently need to meet words often in order to learn them, and coincidentally this is just what natural text offers: the first 2,000 words of English are presented in the language in great frequency. After that, learners somehow manage to learn words from less frequent encounters: and this is just how words after the 2000 mark present themselves, much less frequently. The learning algorithm and the statistics of frequency seem balanced. This is another facet on the ‘learnability’ of language (Pinker, 1989). In summary, our findings support the notion that natural text is designed for lexical acquisition.

However, an important caveat is in order--that natural text is designed for lexical acquisition for native speakers, with 20 years for language learning at their disposal, and not necessarily for second language learners, who normally need to learn a second language in much less time.

The results of this study revealed that approximately one word was learned for every fourteen tested (i.e. 2.16 out of 30). As already mentioned, these findings replicate and confirm those of other incidental vocabulary acquisition research. Thus, there appears to be some measure of regularity to incidental vocabulary acquisition through reading. However, the learning rate established in this type of experiment is often interpreted to mean that reading should be an effective way for learners to develop their lexical knowledge. Is this really plausible? The average number of words learned across all groups was 2.16 out of 30, and the figure for the lower level learners was 1.82 out of 30. The text was 2098 words in length. Thus, less than one word was learned per 1000 words read. Will this learning rate allow a student scoring 50% on the 0-2000 level to learn approximately 2000 more words, and reach the 3000 mark indicated by Laufer (1988) as the minimal threshold to reading unsimplified texts with ease and comprehension?

            To answer this question, we must first determine how many words a student might read in a school year. Fielding, Wilson and Anderson (1986) have estimated that grade-five L1 students read approximately 700,000 words per year. From similar figures, Nagy et al. (1985) state that “the number of new words the typical middle-grade child learns in a year is between 750 and 5,500” (p. 250). However, these numbers cannot be transposed to ESL learners in the Quebec high school system. As indicated above, the Quebec Ministry of Education recommends 100 hours of English instruction per year, for a school year of roughly 40 weeks. This works out to two to three periods per week. If at least one of those periods is devoted exclusively to reading, and the story is similar to the one used for this experiment (between 1000 and 3000 words in length), then students should be acquiring roughly two new words per week incidentally through reading. In a school year of roughly 40 weeks, students would acquire only 70 new words a year, and at this rate 2,000 new words would be learned in just under 29 years.

Of course, it can be argued that vocabulary acquisition does not occur only through reading. Through communicative language teaching, with its emphasis on communicating messages and creating meaningful contexts in which to acquire language, vocabulary learning may take place through many other types of learning experiences in the classroom. What is the extent of vocabulary input that occurs in communicative ESL classrooms? In a study which examined vocabulary input in several primary and secondary school ESL classrooms in Quebec, Lightbown, Meara, and Halter (1999) found few differences in the quantity and quality of vocabulary input between classes where the input was restricted via an audio-lingual method and where the students received intensive communicative instruction. Therefore, the assumption that incidental vocabulary acquisition will naturally occur through communicative teaching cannot be made.

It can also be argued that our own frequency findings have shown that the number of occurrences needed for acquisition drops considerably over the course of learning the second thousand words; in other words, the learning rate is unlikely to remain static but rather to increase. How much this rate could be expected to increase is the most important new question to emerge from the present study. However, even if we quadrupled the figure of 70 words to account for rate increases and vocabulary acquisition that might take place through reading outside the classroom, 280 words per year would still be a low figure (and almost exactly one half the figure of 550 words cited above from Milton and Meara's, 1995, British study). Even at this rate, learning 2000 new word families would take these students more than seven years, and of course the Grade 7 students in this study have only another four years of school before them. One might argue that these statistical extrapolations are somewhat fanciful; however, research by Cobb (in press) with Francophone students enrolled in ESL courses in a French-speaking Quebec university shows that the average vocabulary level of a graduate of the Quebec school system is closely in line with them.

            Readers of this study who work in schools with different arrangements from those in Quebec should work out their own extrapolations from the number of words their students read and the number of hours of ESL instruction they receive. But for Quebec learners, the main finding of our paper seems clear: school learners who knows about 1,000 words of English, roughly the position of many if not most of them, will not establish a minimal functional lexicon through the reading they do in school.

What is the alternative to natural acquisition through reading? We would argue that acquisition through reading be supplemented by either direct vocabulary instruction (Nation and Waring, 1997) or instructionally enhanced reading (Hulstijn et al, 1996). Direct instruction has often been viewed as taboo in the communicative era, but in fact given the rates and predictions offered above, it is entirely reasonable to argue that some form of explicit instruction will be needed to get many students over the 3000-word threshold where they have some chance of reading texts independently and beginning to acquire any significant amount of vocabulary on their own. In most school subjects, the case for an element of direct instruction would be uncontroversial. For example, in geometry, it is quite possible that students could incidentally learn a little geometry simply by observing the shapes and forms of the world around them, but without a basic set of enabling concepts—without the conditions of learning in place—it is unlikely they would ever stumble upon the Pythagorean theorem. In the case of reading and vocabulary, the enabling conditions for reliable comprehension and further contextual learning have been tentatively defined as 3000 word families. Much research into acquisition of syntax within a communicative regime has made a similar point--certain aspects of L2 learners’ grammatical development do not go beyond a minimum level unless some effort is made to draw the learners’ attention to language forms (e.g. Ellis, 1995; Harley, 1994; Norris & Ortega, 2000; Spada, 1997; several contributors to Doughty & Williams, 1998). Similar results have been found in some of the recent work on the contributions of explicit instruction for vocabulary acquisition (Ellis, 1994; Hulstijn, in press; Zimmerman, 1997).

There is no reason that direct or enhanced instruction, whether in syntax or vocabulary, needs to be boring. There are numerous methods for teaching vocabulary outside of the context of reading, ranging from word puzzles to communicative games. With the advent of the computer in classrooms there are a number of programs designed to develop vocabulary that students find quite enjoyable. For example, the Université du Québec à Montréal’s Compleat Lexical Tutor (Cobb, 2000) provides a learning environment in which learners can work their way through the various frequency lists in a number of seemingly pleasurable activities in which contextual inference and frequency based content selection are built into a single learning system. The idea of instructionally enhanced reading also shows great promise. Users of The Lexical Tutor can, for example, read and listen to extended texts, click on words for dictionary information or for concordances displaying words in several additional contexts, and they can record their lexical findings in a database--in some version of what lexicographers do (Cobb, 1999). To be sure, the idea of direct instruction is not to replace contextual learning with vocabulary drills, but rather to provide the conditions, as best as these can be defined at present, to make contextual learning more effective.

 

Notes

 

1. The basis for Laufer’s (1989) definition of 'functional lexicon' is her finding that reading comprehension tends to become reliably functional for academic reading when 90-95% of individual words in a text are known, and that this commonly occurs when learners knows the most frequent 3,000 word families (words and their most usual derivations and inflections). There also exist both variants on Laufer's idea (Sutarsyah, Nation & Kennedy, 1994, propose a lexicon of about 3000 words but chosen for domain as well as general frequency), and alternatives to it in other languages (Hazenberg & Hulstijn, 1996, argue that 11,000 word families are needed for reading Dutch).

 

2. This represents 25 hours less than the MEQ stipulates. The school’s administration claims that this deficiency is compensated for by the fact that ESL classes are divided into levels of student ability, whereas in most public schools, students of varying English language proficiency are put into the same groups.

 

3: This test is effectively the standard English vocabulary test (Meara, 1996, p. 138), despite some reliability problems stemming from the non-independence of items, and other concerns raised by Cobb (in press) and Beglar and Hunt (1999).

 

4. Fewer participants completed the pretest than the initial test. The reason for this was unrelated to the study but due to absences during testing administrations.

 

References

Beck, I. L., McKeown, M. G., & McCaslin, E. S. (1983). Vocabulary development: All contexts are not created equal. Elementary School Journal, 83, 177-181.

Beglar, D., & Hunt, A. (1999). Revising and validating the 2000 word level and university word level vocabulary tests. Language Testing 16 (2), 131-162.

           Beheydt, L. (1990). CALL and vocabulary acquisition in Dutch. In P.J. Kingston, C. Zaehner & A. Beutner (Eds.) Languages, Continuity, Opportunity. London: CILT.

Brown, C. (1993). Factors affecting the acquisition of vocabulary: Frequency and saliency of words. In Huckin, T., Haynes, M. & Coady, J. (1993). Second language reading & vocabulary learning (pp. 263-286). Norwood, NJ: Ablex Publishing.

Cobb, T. (In press.) Vocabulary size tests: One size fits all? Accepted by Canadian Modern Language Review, August 2000.

Cobb, T. (2000).The compleat lexical tutor for data-driven learning on the web. [Computer program]. Montreal: Université du Québec. Available at http://132.208.224.131/.

Cobb, T.M. (1999). Applying constructivism: A test for the learner-as-scientist. Educational Technology Research and Development, 47 (3), 15-31.

Cobb, T. (1998). HyperVocabProfile. [Computer program]. University of Quebec at Montreal, as adapted from Hwang and Nation (1994).

Doughty, C. & Williams, J. (Eds.) (1998). Focus on form in classroom second language acquisition. Cambridge: Cambridge University Press.

Ellis, N. (1994). Consciousness in second language learning: Psychological perspectives on the role of conscious processes in vocabulary acquisition. In J. Hulstijn & R. Schmidt (Eds.), AILA Review, 11, 37-56.

Ellis, N. (1995). Consciousness in second language acquisition: A review of field studies and laboratory experiments. Language Awareness, 4:3 123-146

Fielding, L., Wilson, P., & Anderson, R. (1986). A new focus on free reading: The role of trade books in reading. In T.E. Raphael & Reynolds (Eds.), Contexts of Literacy. New York: Longman.

Harley, B. (1994) Appealing to consciousness in the L2 classroom. AILA Review, 11, 57-68.

Harley, B. (1996) Vocabulary learning and teaching in a second language [Special issue.] The Canadian Modern Language Review, 53 (1).

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

Herman, P.A., Anderson, R.C., Pearson, P.D., & Nagy, W.E. (1987). Incidental acquisition of word meaning from expositions with varied text features. Reading Research Quarterly, 22 (3), 263-284.

Horst, M. (2000). Horst, M. (2000). Text encounters of the frequent kind: Learning L2 vocabulary through reading.  Unpublished doctoral dissertation, University of Wales, Swansea.

Horst, M., Cobb, T., & Meara, P. (1998). Beyond A Clockwork Orange: Acquiring second language vocabulary through reading. Reading in a Foreign Language, 11 (2), 207-223.

Huckin, T., Haynes, M. & Coady, J. (1993). Second language reading & vocabulary learning. Norwood, NJ: Ablex Publishing.

Hulstijn, J. H. (1992). Retention of inferred and given word meanings: Experiments in incidental vocabulary learning. In P. Arnaud & H. Béjoint (Eds.), Vocabulary and applied linguistics. London: MacMillan.

Hulstijn, J. H. (in press). Intentional and incidental second-language vocabulary learning: A reappraisal of elaboration, rehearsal and automaticity. In P. Robinson (Ed.), Cognition and Second Language Instruction. (pp.1-21) Cambridge University Press.

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

Hwang, K., & Nation, P. (1994). Vocabprofile. [Computer program]. English Language Institute, Victoria University of Wellington, New Zealand.

Jenkins, J.R., Stein, M.L., & Wysocki, K. (1984). Learning vocabulary through reading. American Educational Research Journal, 21 (4), 767-787.

Krashen, S. (1989). We acquire vocabulary and spelling by reading: Additional evidence for the input hypothesis. The Modern language Journal, 73, iv, 439-464.

Laufer, B. (1988). What percentage of text-lexis is essential for comprehension? In C. Lauren and M. Nordman (Eds.) Special language: From humans thinking to thinking machines. Clevedon: Multilingual Matters.

Lightbown, P.M., Meara, P. & Halter, R. (1999). Contrasting patterns in classroom lexical environments. In D. Albrechtsen, B. Henriksen, I. Mees, & E. Poulsen (Eds.) Perspectives on foreign and second language pedagogy. Odense: Odense University Press.

Meara, P. (1993). Tintin and the world service: A look at lexical environments. IATEFL: Annual Conference Report, 32-37.

Meara, P. (1995). Single-subject studies of lexical acquisition. [Special issue.] Second Language Research, 11 (2).

Meara, P. (1996). The dimensions of lexical competence. In G. Brown, K. Malmkjaer, & J. Williams (Eds.), Performance and competence in second language acquisition. Cambridge: Cambridge University Press.

Milton, J., & Meara, P. (1995). How periods abroad affect vocabulary growth in a foreign language. ITL Review of Applied Linguistics, 107/108, 17-34.

Mondria, J-A., & Wit-De Boer, M. (1991). Guessability and the retention of words in a foreign language. Applied Linguistics, 12 (3), 249-263.

Nagy, W. E. Herman, P. A. and Anderson, R. C. (1985). Learning words from context. Reading research Quarterly, XX/2, 233-253.

Nagy, W.E. (1997.) On the role of context in first- and second-language vocabulary learning. In Schmitt, N. & McCarthy, M. (Eds.) Vocabulary: Description, acquisition, pedagogy (pp. 64-83). Cambridge: Cambridge University Press.

Nation, P. (1982). Beginning to learn a foreign vocabulary: A review of the research. RELC Journal, 13 (1), 14-36.

Nation, P. (1990). Teaching & Learning Vocabulary. Rowley, MA: Newbury House.

Nation, P., & Waring, R. (1997). Vocabulary size, text coverage, and word lists. In Schmitt, N. & McCarthy, M. (Eds.) Vocabulary: Description, acquisition, pedagogy (pp. 6-19). Cambridge: Cambridge University Press.

Norris, J. M. & Ortega, L. (2000). Effectiveness of L2 Instruction: A research synthesis and quantitative meta-analysis. Language Learning, 50: in press.

Parry, K. (1991). Building a vocabulary through academic reading. TESOL Quarterly, 25 (4), 629 - 653.

Pinker, S. (1989). Learnability and cognition: The acquisition of argument structure. Cambridge MA: MIT Press.

Rott, S. (1999). The effect of exposure frequency on intermediate language learners' incidental vocabulary acquisition and retention through reading. Studies in Second Language Acquisition, 21, 589-619.

Saragi, T., Nation, P., & Meister, G. (1978). Vocabulary learning and reading. System, 6, 72-80.

Schmitt, N. (2000). Vocabulary in language teaching. Cambridge: Cambridge University Press.

Schmitt, N. & McCarthy, M. (Eds.) (1997). Vocabulary: Description, acquisition, pedagogy. Cambridge: Cambridge University Press.

Schouten-van Parreren, C. (1989). Vocabulary learning through reading: which conditions should be met when presenting words in texts? In P. Nation & R. Carter (Eds.), Vocabulary acquisition. Amsterdam: Free University Press (AILA Review/Revue de l’AILA 6).

Singleton, D. (1997). Learning and processing L2 vocabulary: State of the art article. Language Teaching 30, 213-225.

Singleton, D. (1999). Exploring the second language mental lexicon. Cambridge: Cambridge University Press.

Shefelbine, J. L. (1990). Student factors related to variability in learning word meanings from context. Journal of Reading Behavior, Volume XXII, No. 1, 71-97.

Spada, N. (1997). Form-focused instruction and second language acquisition: A review of classroom and laboratory research. Language Teaching, 30, 73-87.

Sutarsyah, C., Nation, P. & Kennedy, G. (1994). How useful is EAP vocabulary for ESP? A corpus based study. RELC Journal, 25 (2), 34-50.

VARGA (Vocabulary Acquisition Research Group Archive, University of Wales, Sansea). [On-line]. Available: http://www.swan.ac.uk/cals/calsres/varga/index.htm .

Wesche, M. & Paribakht, S. (Eds.) (1999) Incidental L2 vocabulary acquisition: Theory, current research, and instructional implications. New York: Cambridge University Press (Special issue of Studies in Second Language Acquisition 21, 2).

Woods, A., Fletcher, P., & Hughes, A. (1986). Statistics in language studies. Cambridge: Cambridge University Press.

Zimmerman, C. (1997). Do reading and interactive vocabulary instruction make a difference? An empirical study. TESOL Quarterly, 31 (1), 121-140.

Zimmerman, M. (1988). Texas indexer/browser, v. 0.27. [Computer program]. Silver Spring, Maryland.

Appendix 1

Items from two levels of Nation’s Levels Test

Testees are asked to identify the meanings of three of the words on the left, by writing the number of the appropriate word beside the given meaning. Item (a) is taken from the 2000 level, item (b) from the University Word List level.

 

(a)    1. blame

        2. hide

        3. hit

        4. invite

        5. pour

        6. spoil

 

 

___ keep out of sight

___ have a bad effect

___ ask

 

 

 

(b)    1. affluence

        2. axis

        3. episode

        4. innovation

        5. precise

        6. tissue

 

 

___ introduction of a new thing

___ one event in a series

___ wealth

 

 

 

 

Appendix 2

Text: The Golden Fleece

 

Myths are stories that show what people believed long ago. The ancient Greeks believed the world was ruled by many different gods. They believed that most events were caused by these gods.

The Golden Fleece is a Greek myth. It shows that the ancient Greeks believed in magical events. They believed that certain people, called oracles, could see into the future. In The Golden Fleece, an oracle predicts the future for a young hero named Jason. This is the kind of myth we would call an adventure story today. Jason faces many dangers as he tries to win his kingdom from his evil uncle, Pelias. Jason's hardest task is to capture the golden wool of a magic sheep. This wool is called the Golden Fleece. It is guarded by a dragon in a distant land.

Can Jason fight the dragon and win? Can he prove his strength, courage, and wisdom?

Old King Aeson of Thessaly was worn and weary with the heavy cares of governing his people. He longed to have his handsome son, Jason, take his place on the throne. But Jason was young, much too young to rule over a kingdom. Yet the king was too old and too tired to go on.

So Aeson called his brother, Pelias, before him and said: "My brother, I can no longer rule. While my little son, Jason, is still a child, I would have you take the throne in my place. But you must agree that when Jason reaches manhood, he, the rightful ruler, shall be king.”

Pelias agreed. But secretly he made up his mind that Jason would never be seated upon the throne of Thessaly. To find out just how he might keep the kingdom for himself, Pelias went to an oracle. The oracles were fortune-tellers. In those days, people believed that an oracle had the power to see into the future.

After mumbling a few charms and burning some magic powders, the oracle told Pelias, "Don't fear anyone who doesn't come with and without a shoe!"

This strange reply satisfied Pelias that he had nothing to fear from Jason. But he wanted to make sure that Jason would never cause him trouble. After the old King Aeson died, Pelias had Jason banished to a faraway land, many weeks' journey from Thessaly.

 There in that far-off country, Jason was reared. He grew from babyhood into boyhood; and when it came time for Jason to go to school, he was sent to study with the centaurs. A centaur was a strange creature, half man and half horse. Below the waist, the centaur had the four legs and body of a horse, which gave him the speed and strength of that animal. From the waist up, the centaur was a man, gifted with intelligence. The centaurs were famous for being very learned. In addition, they were known for their tremendous bravery and skill as fighters and archers.

Young Jason's special teacher was Chiron, the chief of the centaurs. Chiron was delighted with Jason's wit and quickness. Chiron taught so well, and Jason learned so readily, that soon the youth was a master of the manly arts and had a goodly store of the world's wisdom.

Jason did not know that he was a king's son. But when the young prince reached manhood, Chiron the centaur was satisfied that Jason was fit to be a king. He told Jason the story of his royal birth. Jason decided to win his throne back from his wicked uncle, Pelias.

So Chiron asked an oracle for advice on how Jason should overcome his uncle. Again came a strange reply: "Who seeks the crown shall wear a leopard's hide,” said the oracle.

Following the oracle's words, Jason went deep into the forest and killed a leopard. Then Jason dressed himself in the leopard's skin. He said farewell to Chiron and set forth on his way to Thessaly.

Soon Jason came to a rushing river. The stream was wide and rough. Its current was so strong that the people who stood on the river bank warned Jason not to try to cross. Many had tried, the people said, but no one had been able to cross the river alive.

But Jason stepped into the raging current. The swirling waters pulled angrily at the young man who dared its power. The trunks of huge trees being carried down the stream rushed at Jason like mad things. One slip - one moment of carelessness - and Jason would go down, never to be seen again. Step by step he made his way across. Those who watched from the river's banks held their breaths. There were only three steps, two steps, one step more to go - when suddenly the watchers saw Jason swept off his feet!

A groan went up from many lips. The young man who had braved the current was lost! But no! One powerful arm reached up and grasped a large branch of a tree that hung over the river. Slowly Jason pulled himself to safety on the far bank. He had won his first great test!

Jason rested for a moment. Then he stood up. Suddenly he realized that one of his feet was bare. He had lost one of his sandals in the rushing river. But Jason wanted to reach Thessaly, so he strode forward.

It was not long before Jason came to his own city. As he walked through the streets of Thessaly dressed in a leopard skin and one sandal, there was a great commotion. Seeing this handsome youth, people remembered the old prophecy that a young man would appear wearing only one sandal, and that he would be their rightful king.

The false king pretended that he was glad to see Jason. Pelias said he had only been holding Jason's kingdom for him until his return and that he would be glad now to turn the throne over to the returning prince.

Then Pelias tried to trick Jason with a royal feast of welcome. With his nephew in fine humor, Pelias craftily told Jason of a glorious adventure. It would make the youth famous all over the world. "Far away, in the land of Colchis, there is a Golden Fleece," he said. "This fleece is the wool of a golden sheep that is the rightful property of your family. But the Golden Fleece is held by the king of Colchis. It is guarded by a dragon that never sleeps. If I were a young man, nothing would stop me from winning the Golden Fleece."

 When Jason heard this story, he was determined to win fame and glory by capturing the Golden Fleece. This was just as Pelias had planned. He was sure that Jason would be killed.

Jason called for brave men to accompany him on his adventure. From all parts of Greece they came. Jason chose only the bravest and the best. Soon there were 50 of the greatest heroes of Greece together in Thessaly.

Jason asked Argus, a famous boat' builder, to build him a ship large enough to hold 50 men. So Argus built a vessel, the likes of which had never been seen, before. It was strong and large.

When their ship was finished, the heroes called it the Argo, after Argus, the builder; and they themselves were called the Argonauts. The 50 heroes set forth from Thessaly with Jason at their head. Good fortune seemed to favor them. For when the Argonauts came to Thrace, they were warned by Phineus, the sage, of a great danger that lay before them.

In the sea ahead, there were two rocky islands between which they must pass. Whenever anything tried to pass, the islands would come crashing together at great speed. Anything caught between the two islands was crushed to splinters.

"To avoid this terrible catastrophe,” said Phineus, "when you approach the islands, let loose a dove. The dove, flying between the islands will cause them to rush together. But the dove can fly through swiftly enough to escape destruction. Then, at the moment when the islands start returning to their former places, you and your men row with all your might. With the help of the gods you may be fast enough to get through before the islands rush together again.

Jason and his companions thanked Phineus and continued on their way. Soon they came to the islands. The Argonauts drew as near to the fearful passage as possible. Then they released the dove. Straight as an arrow the bird flew between the islands. A roaring sound filled their ears as the islands came rushing together with a crash. But the dove was safe; only her tail feathers were caught between the rocky walls of the islands.

Jason's men sat at the oars, ready at Jason's signal to row with all their strength. The moment the islands had separated enough for the Argo to fit through, Jason gave the signal. Then, as one man, the Argonauts pulled their oars. The sweat stood out on every brow as they pulled. Swift as a bird the Argo, with its crew of heroes, sped between the islands. The islands separated their full distance, and then with fearful speed came crashing together again. By just the breadth of a hair, the Argo slipped through. Actually, the tip of the dreadful islands grazed the stern of the boat.

At last the heroes came to the kingdom of Colchis. Jason made known to the king his determination to get the Golden Fleece. The king laughed. "Oho, you think that all you have to do is come here and demand the Golden Fleece and it is yours! Not so fast, my young fellow. The first thing you must do is to harness the fire-breathing bulls. Then you must plow the stony field that lies next to the temple of Mars. After that, you must sow the dragon's teeth in the field.”

But Jason was not frightened or discouraged. "Let us not delay a moment," he cried. "Lead us to our tasks!"

So the king led Jason and his companions to the field of the bulls. The fierce red eyes of the bulls glared angrily at Jason and his bold crew. As the bulls pawed the ground, the fiery breath curled from their nostrils and scorched the ground for many yards around them. No one could get close to the ferocious beasts without being burned to a crisp by the flames they breathed. The Argonauts were forced to withdraw from the field.

That night, in the Argonauts' camp, Jason was in despair. Suddenly a maiden stood before him. It was the king's daughter, Medea. She had fallen in love with Jason.

"Do not be downhearted," she said, "No mortal man could have overcome those enchanted bulls without help. But I can help you. If you promise to marry me, I will help you to win the Golden Fleece.”

Jason agreed. Then she gave him a magic herb. "This will tame the bulls," she said. Next, she gave him a heavy black stone, saying, "Use this when the need arises." At dawn the next morning, Jason went into the field alone. When the bulls came stamping toward him, he held out the magic herb and lo! the bulls became tame! They permitted Jason to yoke them to the plow.

When the king and his court awoke, there was Jason, quietly plowing the field of Mars with the now-gentle bulls.

The king was furious. He suspected treachery, but he never suspected his own daughter.

"Here," he said, "in this warrior's helmet are the dragon's teeth. Sow them in the field you have plowed." Jason took the helmet and did as he was told. No sooner had he dropped the seeds than little shiny points began to appear. As Jason watched, the points grew. Soon, to his horror, he saw a most remarkable thing. Each of the teeth was becoming an armed warrior. First their spears appeared, then their helmets and, right under his eyes, Jason saw an army of soldiers come up out of the ground. No sooner had they sprouted than the soldiers and their spears rushed upon him. Jason drew his sword to fight, but how could one man hope to win against so many?

Then he remembered the weapon Medea had given him. The stone. That was it! Now must be the time to use it!

Quick as a flash, Jason threw the stone right into the middle of the army. The man who was hit, sure that his neighbor had struck him, turned angrily and struck him back. In a matter of moments, the thousand soldiers who had sprung from the dragon's teeth were fighting each other. They fought each other, instead of Jason, until not one of the army was left alive.

Now the king's daughter quickly drew near to Jason. "Quick!" whispered Medea. "To the fleece!”

She handed Jason a magic potion. Jason rushed to the garden where the fleece, guarded by the dragon who never slept, was hanging on a tree.

Jason sprinkled a few drops of the potion before the dragon. Its hideous head rolled from side to side as it fell into a deep sleep. Then Jason drew his sword and cut off the dragon's head. He looked about him, and there hanging on the limb of a tree was the Golden Fleece. It gleamed like the sun. Jason captured the precious prize and shouted to his companions to follow him to their ship. Medea ran for the ship, too, for she knew what her fate would be when her father discovered her part in helping Jason - as he surely would.

 In a rage, the king gathered his men together. He chased his daughter and the Argonauts, hoping to catch them before they reached their ship. But the Argonauts were too swift and too strong for the king's men, and they made their escape. The Argo, with its brave crew, had many dangerous moments on the return voyage. But with the help of the gods, who applauded their bravery, they at last reached Thessaly again.

Now, with the help of Medea and the 50 heroes, Jason forced the evil King Pelias to yield the kingdom to him, its rightful ruler.

Appendix 3

Pre-post test

Name

Group

Grade

Put the number of the word in the left column in the space beside the correct definition .

 

1.   helped

2.   drew

3.   rightful

4.   vein

5.   even

6.   stream

 

 

___a small river

___pulled

___legal

 

1.   fearful

2.   kindness

3.   freedom

4.   wisdom

5.   timid

6.   ruler

 

 

___leader

___terrible

___intelligence

 

 

1.   spew

2.   rush

3.   youth

4.   field

5.   tough

6.   push

 

 

___to go with speed

___a young man

___area of land

 

1.   sheep

2.   moats

3.   hungry

4.   entaiI

5.   fight

6.   angrily

 

 

___with rage
___battle

___grass-eating animals

 

 

1.   aspire

2.   sword

3.   breath

4.   manhood

5.   homily

6.   glove

 

 

___weapon with a long thin blade

___air brought into and out of the mouth

___the age of an adult male

 

1.   fleece

2.   toys

3.   rams

4.   dent

5.   oars

6.   teeth

 

 

___ used for biting and chewing

­­___the thick hair of an animal

___used to row a boat

 

1.   crash

2.   brew

3.   laughter

4.   bath

5.   daughter

6.   sow

 

 

___to plant

___someone's female child

___come together violently

 

 

 

1.   chevron

2.   forth

3.   plow

4.   decant

5.   centaur

6.   chart

 

 

___ to dig the soil

___half man half horse

___onwards, out from home

 

 

 

1.   lands

2.   marks

3.   cast

4.   helmet

5.   crew

6.   spears

 

 

___weapons that are thrown

___people working on a ship

___hat worn for protection

 

 

1.   dove

2.   oracle

3.   column

4.   swift

5.   vital

6.   visor

 

 

___person who can see into the future

___a white pigeon-like bird

___quick or rapid