erten&tekin (2008)

26
Effects on vocabulary acquisition of presenting new words in semantic sets versus semantically unrelated sets İsmail Hakkı Erten , a , , and Mustafa Tekin a , a Çanakkale Onsekiz Mart University, Egitim Fakultesi, Ingiliz Dili Egitimi Ana Bilim Dalı, 17100 Çanakkale, Turkey Received 12 November 2007; revised 24 January 2008; accepted 8 February 2008. Available online 17 July 2008. Abstract This paper reports on a study which investigated the effect on vocabulary recall of introducing new words via two different methods. A one-group quasi-experimental research design with alternating time series measures was employed. A group of 60 fourth graders were taught 80 carefully selected words either in semantically related sets or semantically unrelated sets. Also under investigation was the effect of these methods on test completion. The statistical analysis revealed that learning words in semantically unrelated sets yields better results than learning vocabulary in semantically related sets. The difference persisted in the long term. Further, test completion time was much longer for the semantically related vocabulary items, indicating a slower recall of vocabulary. The study indicated that, contrary to frequent practice in many course books, presenting new vocabulary that belongs to the same semantic set together may cause interference due to cross-association and may even hinder vocabulary learning. Such practice needs to be questioned and alternative methods that involve presenting vocabulary in unrelated sets need to be developed to facilitate vocabulary teaching and learning. Keywords: Interference theory; Mental lexicon; Semantic set; Thematic set; Vocabulary recall 1. Introduction

Upload: kiri-wood

Post on 03-Oct-2015

219 views

Category:

Documents


0 download

DESCRIPTION

Effects on vocabulary acquisition of presenting new words in semantic sets versus semantically unrelated sets

TRANSCRIPT

Effects on vocabulary acquisition of presenting new words in semantic sets versus semantically unrelated sets

Effects on vocabulary acquisition of presenting new words in semantic sets versus semantically unrelated sets

smail Hakk Erten

, a,

,

and Mustafa Tekina,

aanakkale Onsekiz Mart University, Egitim Fakultesi, Ingiliz Dili Egitimi Ana Bilim Dal, 17100 anakkale, Turkey

Received 12 November 2007;

revised 24 January 2008;

accepted 8 February 2008.

Available online 17 July 2008.

Abstract

This paper reports on a study which investigated the effect on vocabulary recall of introducing new words via two different methods. A one-group quasi-experimental research design with alternating time series measures was employed. A group of 60 fourth graders were taught 80 carefully selected words either in semantically related sets or semantically unrelated sets. Also under investigation was the effect of these methods on test completion. The statistical analysis revealed that learning words in semantically unrelated sets yields better results than learning vocabulary in semantically related sets. The difference persisted in the long term. Further, test completion time was much longer for the semantically related vocabulary items, indicating a slower recall of vocabulary. The study indicated that, contrary to frequent practice in many course books, presenting new vocabulary that belongs to the same semantic set together may cause interference due to cross-association and may even hinder vocabulary learning. Such practice needs to be questioned and alternative methods that involve presenting vocabulary in unrelated sets need to be developed to facilitate vocabulary teaching and learning.

Keywords: Interference theory; Mental lexicon; Semantic set; Thematic set; Vocabulary recall

1. Introduction

Vocabulary has lately gained popularity in the general field of English language teaching and learning and become a guest of honour ([Coady and Huckin, 1997], [Read, 2000], [Richards and Renandya, 2002] and [Bogaards and Laufer, 2004]) with a wide range of research and pedagogical interest. Of the many dimensions of research in this field, there does not seem to be a consensus on several issues, with some controversial results and methodological bias. One such issue concerns whether new vocabulary should be presented in semantically related sets or semantically unrelated sets. This paper aims to address this controversy. At one end of this conflict, there are authors like (Seal, 1991), (Grandy, 1992), (Haycraft, 1993), (Stoller and Grabe, 1995) and (Wharton and Race, 1999), and Hashemi and Gowdasiaei (2005), who speak in favour of presenting new words in semantic sets on the basis that it is an effective way of presenting new words, and possibly reflecting the natural organization of the mental lexicon ([Aitchison, 1994] and [Aitchison, 1996]). On the other hand, there are those researchers ([Higa, 1963], [Laufer, 1989], [Tinkham, 1993], [Tinkham, 1997], [Waring, 1997], [Nation, 2000] and [Finkbeiner and Nicol, 2003]), who maintain that if similar words that share numerous common elements and a super-ordinate concept are introduced at the same time, these words will interfere with each other and have a negative effect on their retention due to cross-association and possible overloading in the short term memory.

1.1. Arguments FOR presenting vocabulary in semantic sets

Using word associations in vocabulary teaching has gained currency over several decades and has found its way into materials (McCarthy, 1990). In fact, a brief examination of some course books that are commonly used in English lessons reveals that many of them tend to introduce new words in semantic sets (e.g. New Headway Elementary Soars and Soars, 2000; New English File Oxenden et al., 2004). According to Waring (1997), this common practice in vocabulary instruction derives from a pervasive belief among course book writers that presenting new vocabulary in semantically related groups will facilitate vocabulary building. He further points out that this belief is not founded on research but on methodology and convenience.Numerous SLA theorists and practitioners defend (implicitly or explicitly) the position that teaching new L2 vocabulary in semantically grouped sets is an effective method of vocabulary instruction. Much of the support for this position comes from studies of the organization of the mental lexicon in L1. They claim that this method is in compliance with various brain theories which suggest that there is a good organization of semantic fields in the human brain ([Carter and McCarthy, 1988], [McCarthy, 1990], [Grandy, 1992], [Aitchison, 1994], [Rogers, 1996] and [Lewis, 1997]). Furthermore, evidence has been provided for the view that words are semantically organized in the human brain and that individuals tend to recall words on the basis of the semantic field in which they are conceptually mapped ([Aitchison, 1994] and [Aitchison, 1996]). Therefore, it is held that teaching a large number of words in an unrelated way can be likened to imagining a tree with no trunk and branches, but only leaves (Haycraft, 1993). Haycraft goes on to maintain that it is easier to teach vocabulary items that belong to the same semantic field because the learner will be able to form a pattern of interrelated words in his mind. This is because such an approach would have several advantages, one of which is that by learning items in sets, the learning of one item can be reinforced by the learning of another ([Seal, 1991] and [Wharton and Race, 1999]). Recently, Hashemi and Gowdasiaei (2005), reflecting on their own findings, assert that vocabulary should be presented in semantic sets within an appropriate context as learners will get involved in deeper levels of mental processing (Craik and Lockhart, 1972). Thus students make more effort to differentiate words that belong to the same semantic set and their learning will endure longer.

1.2. Arguments AGAINST presenting vocabulary in semantic sets

It is often contended nowadays that the belief that presenting new vocabulary in semantic sets facilitates learning is a common vocabulary myth unsupported by sound research results ([Waring, 1997] and [Folse, 2004]). In fact, research as early as 1931 revealed that when several similar words were introduced at the same time, it had an interfering effect on learning and that especially synonyms were learned very poorly by the participants (McGeoch and McDonald, 1931). This is often explained by referring to Interference Theory ([Baddeley, 1997], [Anderson, 2003] and [Reed, 2004]).Convincing evidence has recently been offered for this position ([Tinkham, 1993], [Tinkham, 1997], [Waring, 1997] and [Finkbeiner and Nicol, 2003]). Tinkham, in his experiments, found that learning new words grouped in semantic sets required more learning trials to be learnt completely. Waring confirmed Tinkhams findings. Further, Finkbeiner and Nicols experiment illustrated that recall of words learnt in semantically related sets is slower in both directions of translation (L1L2 or L2L1). Finally, a recent study by Pigada and Schmitt (2006) revealed that words confused by learners are not only the semantically related ones but words that have similar forms (Laufer, 1997). This is probably because it takes longer to differentiate and to assign new labels for new words in semantically related items (Nation, 2000). These findings can be translated into an assertion which claims that semantically similar new words might have a deleterious effect on learning (Finkbeiner and Nicol, 2003, p. 376) and actually impede rather than facilitate the learning of new vocabulary items (Tinkham, 1993).It should also be noted, however, that there is a difference between the presentation of new words in semantic sets and presenting them in thematic sets ([Tinkham, 1997] and [Waring, 1997]). According to Waring, it is not possible to prevent all semantically related words from appearing in coursebooks. In fact, a list of words in a particular unit for some learners may trigger recall from previous learning, and for other learners, it may constitute a list of new words to be learned. Therefore, Tinkham and Waring advise a thematic rather than a semantic arrangement of new vocabulary items because a thematic group which contains the words sweater, changing room, try on, wool, striped may not cause as much confusion as the words scarf, tie, coat, pants and skirt, which constitute a semantically related set. The words in a thematic group are all related even though they do not form a proper semantic set with a super-ordinate term and co-hyponyms.A review of research and arguments for and against presenting new vocabulary in semantic sets is far from producing any decisive conclusions. There are conflicting views and different research results. As Hashemi and Gowdasiaei (2005) suggest, further research is needed to elucidate this inconsistency. This study aims at contributing to our understanding of the phenomenon.The study is different from previous studies ([Tinkham, 1993], [Tinkham, 1997], [Waring, 1997] and [Finkbeiner and Nicol, 2003]) that focused mainly on the acquisition of artificial labels to L1 words. In those studies, words in the learners L1 were paired with artificial words from an alien language (e.g., mouse-kunop; pear-okess) and oral repetition was used to present and practice them (Waring, 1997). They then compared the number of trials their subjects made to learn semantically related words with the trials they made to learn unrelated words. Finkbeiner and Nicols (2003) study also involved real word- artificial word pairing and investigated the length of time required for the completion of the learning task. Although these studies shed light on the effects of semantic relationship on the acquisition and recall of lexical items, what is perhaps needed is studies that endeavour to explore the phenomenon with real words. One such study (Hashemi and Gowdasiaei, 2005) reports results contradicting other studies. However, it is difficult to compare their findings as previous studies were mainly concerned with the length of time for completion of the learning task with a set level of success, while this study measures depth and breadth of vocabulary gain (Wesche and Paribakht, 1996). Further, Hashemi and Gowdasiaei do not disclose full details of the list of words they made use of for their experiment, making it difficult to make sense of their findings. Therefore, it is safe to propose that there is room for further research.Another important point with the current study is that, to the best of our knowledge, previous studies of this kind focused mainly on adult learners. This study focuses on young learners, hoping to expand our understanding of the phenomena with different groups of students. Thus, the aim of this study was not to replicate previous studies, but carry out a similar study with young learners learning real English words, hoping to better explore the question under scrutiny.

2. The study

2.1. Aims

The present study examined the effects of presenting new words in semantic sets and in semantically unrelated vocabulary groups without a common super-ordinate concept. The study sought answers to the following research question: Does presenting real English words in semantically related sets versus semantically unrelated sets make a difference in learning?The study also investigated the difference between the length of test completion under the two conditions.

2.2. Methodology

A one-group quasi-experimental research model with an alternating time series design was employed (Hatch and Lazaraton, 1991, pp. 9293). This model, within a time series model, allows measuring the effect of two methods on the same group of students. The effect can be measured by alternating the periods when one set of materials is used with the times when another set of materials is used. In this particular study, 60 students were presented with 80 new English words in two different sets of instructional methods. The vocabulary items selected following certain rigid criteria were divided into four 20-word sets: two semantically related and two semantically unrelated sets. All the students were taught all the words and were tested separately for related and unrelated items. Such an experimental design allowed researchers, within ecological validity (Hashemi and Gowdasiaei, 2005), (a) to reduce the bias that might have been posed by non-equivalence of different groups of students (i.e., personality, learning styles, group dynamics, etc.) and (b) to control the lexical difficulty of both the semantic sets and the unrelated sets, and thus foster internal validity (Trochim and Land, 1982).

2.3. Setting

The study was carried out at a state primary school in western Turkey. One reason why this particular school was chosen was that the school received students from a neighbourhood where mostly middle-class working families lived. It was seen as important to ensure that the participants came from similar socio-economic backgrounds. Working with a homogeneous group would also minimize possible student variations. The school was located centrally, which contributed to homogeneity.The class sizes were not very large and the students had not taken any English courses before the year the study was conducted. Another factor was that the second author was employed as a teacher of English at this school when the research was carried out, which was assumed to be appropriate to ensure natural group dynamics.

2.4. Participants

The study was carried out with two intact groups of participants, all of whom were fourth grade students. At the time of data collection, they had had very limited formal language instruction. Therefore, their vocabulary size was relatively small and they were very similar to each other in terms of their level of English proficiency. This facilitated the selection of vocabulary items to control lexical variables that might have biased the data.Initially, there were 60 participants in two groups. 29 were female and 31 were male students. However, five of the participants were excluded at the analysis stage since they could not attend all sessions.

2.5. Materials and instruments

The vocabulary items used in the study were selected from an initial list of 100 words, where students were asked to match vocabulary items to corresponding pictures. This word list was in the form of picture-word matching to avoid any possible difficulties in comprehension. Of these words, 80 unknown words were selected for the study. In other words, 20 vocabulary items which were correctly guessed by the participants were excluded from the study, leaving 80 words out of an initial list of 100 words.The words included in the experiment were chosen and classified according to their length ([Ellis and Beaton, 1993] and [Erten, 1998]), semantic relations ([Tinkham, 1993], [Tinkham, 1997] and [Waring, 1997]), and cognateness (Nation, 1990), while also considering idiomaticity (Laufer, 1990) and concreteness.The 80 unknown vocabulary items were checked for number of letters as well as number of syllables. Vocabulary items were carefully assigned to different groups and four sets of vocabulary items with equal number of letters and syllables were formed. As shown in Table 1, the semantically related sets had an equal number of letter/word and syllable/word ratio as their opposing unrelated sets.

Table 1. Mean numbers for letters and syllables of words used in the study

WeekLessonsMean number of lettersMean number of syllables

Week 1Lesson 1 (semantic set)3.81.25

Lesson 3 (unrelated set)

Week 2Lesson 2 (unrelated set)4.751.65

Lesson 4 (semantic set)

The semantic set and opposing unrelated set of the first week both had 3.8 letter/word and 1.25 syllable/word ratios while in the second week, the vocabulary items were slightly longer with 4.75 letter/word and 1.65 syllable/word ratios, respectively.Only concrete vocabulary items were chosen for this study since abstract words were considered to be inappropriate for research purposes: they would be difficult to pre-test, illustrate, and measure in terms of recall using visual materials, without which a serious literacy problem could have biased the data. Further, abstract concepts might have been inappropriate for the target age group.Half of the vocabulary items were chosen from two semantic fields. The first semantic set included 20 animal words and the other semantic set contained 20 food words. The other two sets had 20 semantically unrelated vocabulary items each.Another criterion in vocabulary selection involved cognates. No English word that sounded the same as or similar to a Turkish word was chosen for the study, since using cognates might have a facilitating effect ([Nation, 1990] and [Erten, 1998]). Lastly, words did not involve any idiomatic use, in order not to impede the learning process (Laufer, 1990).Such a careful selection and homogenising of vocabulary items, combined with the diminished student variations in a one-group quasi-experimental design suggested by Hatch and Lazaraton (1991), is believed to have formed equally learnable sets of vocabulary items to allow testing of the effect of different methods of vocabulary presentation with fostered internal validity (Trochim and Land, 1982).Flashcards and activity sheets were prepared for the presentation and practice stages of the study. Each flashcard had a big picture of the target word together with its written form under it (see Appendix 2 for a sample flashcard). Each activity handout contained matching type exercises that required the participants to match the written form of the new words on one side with their pictures on the other side. The same pictures were used at all stages of the study to avoid possible problems related to picture identification. Practice-stage activities were also designed in a similar way to the tests used in the study to avoid problems and confusion.Pre-tests were given before each presentation. The same test was repeated as an immediate post-test just after the presentation and practice stages and as a delayed post-test with a reshuffled order to prevent any memory effects as practiced in the field (e.g., Ellis and Beaton, 1993).

2.6. Procedures for data collection

The study was conducted as a part of the normal English course during normal class hours. Therefore, all those present in the class participated in the activities. The new words were presented by means of flashcards and reinforced through repetition.The study as mentioned above was conducted with two groups of students. Both groups of participants were taught all words. Words, however, were presented in different ways to create two different presentations and study conditions as required by the research design.The study was completed within a period of three weeks. During the first week, two sets of words were introduced to the participants in two different class hours on two different days. The first set contained 20 semantically related words (animal words) and this set was presented first. The second set, which was presented two days after the first set, contained various words from different semantic fields as shown in Appendix 1. The following week, it was vice versa: the unrelated set of words was introduced first and the related set (food words) was introduced after the unrelated set.Each presentation began with a pre-test of the target words to be introduced on that day and ended with an immediate post-test to measure the short term recall of vocabulary. If a participant could identify a word correctly, he was considered to have learned that vocabulary item on a binary scale.The lesson/hours lasted 40min. Vocabulary items were presented in a controlled manner in terms of time and actions to provide equal study conditions for each word. A chronometer was used to check the time regularly. In each group and presentation, every effort was made to equalize the number of repetitions, gestures, mimes, and time allocated for each word. The activities used with both groups were the same and included repetition drills and matching words with flashcards. Furthermore, the same materials, namely flashcards and worksheets, were used with both groups.Once the controlled vocabulary presentation was completed, students were instructed to sit silently until the test papers, in which students were asked to match the pictures with the new vocabulary, were distributed. They were told to start answering on the researchers call and to raise their hands when they had completed the test so that the researcher could note the time for test completion. Time was kept by using an electronic chronometer that allowed for continual time keeping, which allowed record of the completion time of all the students using only one chronometer. When more than one student reported completing the test at the same time, their completion time was recorded as such.Considering that the testing was conducted in a natural environment, every effort was made to prevent interaction and possible calling out of answers or even cheating among students during the test. As the second author was the actual class teacher, it was observed that students were all obedient to instructions and no interference was recorded. Table 2 sums up the whole process of the study.

Table 2. Procedures for collecting data

1st Week

Session 1 (Day 1)1st Lesson: related set 1 (animals) (pre-test, presentation, and immediate post-test)

Session 2 (Day 2)Revision of the 1st lesson (flashcards and oral repetition)

Session 3 (Day 3)2nd Lesson: unrelated set 2 (pre-test, presentation, and immediate post-test)

Session 4 (Day 4)Revision of the 2nd lesson (flashcards and oral repetition)

2nd Week

Session 5 (Day 8)3rd Lesson: unrelated set 1 (pre-test, presentation, and immediate post-test)

Session 6 (Day 9)Revision of the 3rd lesson and delayed post-test of the 1st lesson (flashcards and oral repetition)

Session 7 (Day 10)4th Lesson: related set 2 (foods) (pre-test, presentation, and immediate post-test)

Session 8 (Day 11)Revision of the 4th lesson and delayed post-test of the 2nd lesson (flashcards and oral repetition)

3rd Week

Session 9 (Day 18)Delayed post-tests of the 3rd lesson and 4th lesson

2.7. Procedures for data analysis

The research question of whether presenting vocabulary in semantic sets or semantically unrelated sets yields distinctive learning outcomes was investigated by analysing the immediate and delayed post-test results quantitatively. Next, participants completion time of each test was calculated in order to establish whether the type of presentation influenced the length for test completion.

3. Findings and discussion

3.1. Effects of vocabulary presentation

The mean value for each group of words on the pre-test was zero, which indicated sample homogeneity in terms of initial vocabulary knowledge. Therefore, students performance was assumed equal on each set of words. There were no group differences between the two classes on any post-test measures (p