U
NIVERSITEIT
VAN
A
MSTERDAM
The effects of task-induced involvement load and word
exposure frequency on L2 incidental vocabulary learning
through reading
Zuzana van Polen
a thesis submitted to the Graduate School for Humanities, University of Amsterdam in partial fulfilment of the requirements of the degree
Master of Arts (MA) In
General Linguistics 13 June, 2014
Supervisor: Rob Schoonen (UvA)
2 Acknowledgements
I would like to thank to a number of people who have assisted me in the extremely inspiring and rewarding journey of acquiring academic knowledge in second language acquisition as well as in the preparation, the writing process, and editing of this thesis.
First of all, I would like to thank my thesis supervisor, Rob Schoonen, for his astonishingly great expertise, inspiration, and guidance. Rob, thank you so much for the lectures in the second language acquisition that have broaden my knowledge horizon tremendously, that ignited my interest in this field and demonstrated so many wonderful opportunities how to connect applied science and second language education, which has been my work field for some years. Thank you for all the inspiring discussions, your patient guidance and excellent feedback.
In addition, I want to express my gratitude to my students at the University of Applied Sciences in Amsterdam, the Department of Business Administration, who formed the population of the experimental study. For the first time in my professional career as a university lecturer, I was not only teaching them, but had the opportunity and privilege to observe and scientifically explore how they acquired words in English when they read, which led to many revelations. I am indebted to my colleague English lecturers, Tina van Klaren and Louise Vines, who provided me with a great support in the proof-reading of the articles. Thank you both for your plentiful comments and suggestions. I am very grateful to my dear Czech friend, Katerina Chladkova, for her priceless support and guidance in the execution of the statistical analyses. Kata, without you I would have had great difficulties to analyze and interpret the data.
Finally, my very special thanks goes to my dearest husband, Sebastiaan, who has been my cornerstone and has given me the most precious, loving support throughout the two years of my academic endeavour, and even more so during the finalizing phase of my thesis writing. Seb, your encouragement and love have been indispensable to me. And, of course, I want to thank my dear parents living in Slovakia, who have unfailingly supported me in any decisions I took in life, study and work.
3 Abstract
Background: Incidental vocabulary learning refers to word acquisition that emerges as a by-product of listening or reading when the learner’s focus lies on the meaning of language. Research has shown that the pick-up rate of words learned through exposure to them in a reading text is relatively low (Waring & Takaki, 2003). It has therefore been argued that classroom environment cannot fully facilitate the time needed to learn vocabulary incidentally and words that were learned were mostly tested as receptive word knowledge only (Pigada & Schmitt, 2006). The depth of processing as a construct of
Involvement Load Hypothesis (ILH, Hulstijn & Laufer, 2001) is another mediating factor in incidental
vocabulary acquisition. In the research of incidental vocabulary learning through reading involvement load and word exposure frequency were always investigated separately. So far merely one study, Eckerth and Tavakoli (2012), explored these two factors and their impact upon word learning and retention together.
Aims: The current research replicated the study by Eckerth and Tavakoli, aiming to assess the learners’ active word knowledge. Two research questions were formulated: 1.What are the effects of exposure frequency and task-induced involvement load on learners’ two types of word knowledge (active recall and active recognition) in initial word learning and in delayed word retention? and 2.Do word retention rates differ for the two types of word knowledge measured?
Methods and Procedure: In three consecutive weeks 51 Dutch upper-intermediate L2 learners of English were exposed to three reading texts, that each contained 10 TWs occurring either once or 4 times in the article. Involvement load was operationalized through three tasks differing by the depth of mental involvement: Task 1 (reading + glosses), Task 2 (reading + gap filling), and Task 3 (reading + composition writing). To obtain the 30 TWs, learners’ vocabulary knowledge was first pre-tested by a MC recognition test. Every week an immediate word post-test was administered, 4 weeks after the main treatment a delayed post-test took place. Both post-tests assessed active recall and active recognition. Since the data was elicited through a within-group design, multi-factorial repeated measures ANOVA was conducted to analyse the outcomes. A number of post hoc pairwise comparisons of means were carried out to see the differences between the levels of involvement load. Results: Overall, active recognition word knowledge was better acquired and retained than active recall. Both independent variables, involvement load and word exposure frequency, had significant impact upon word learning and retention, but each in a different way. A main effect of involvement upon word knowledge was established, but the task with the highest level of involvement (Task 3) did not lead to highest word gains and retention across both word knowledge types. The analyses yielded an interaction effect between word exposure frequency and active recall word knowledge at both times of post-testing. Active recognition knowledge was better learned and retained than active recall, and there was hardly any loss of this knowledge.
Conclusions: The findings indicate that each of the two factors, task-induced involvement load and word exposure frequency, affects incidental word learning and retention from reading. As for the involvement load, no conclusive support has been lent to ILH, since although strong level of involvement facilitated the active knowledge learning, it cannot be claimed that an input-output task (Task 3) is the most conductive in this respect. It also appears that the word gains obtained initially after completion of Task 3 were not preserved over time. The interaction effect of word exposure frequency was only manifested on active recall word knowledge, meaning that learners could equally well recognize high exposure frequency word forms straight after the main treatment and four weeks later, but they could not recall the word forms equally well.
Keywords: incidental word learning, involvement load hypothesis, word exposure frequency,
4 Table of Contents Acknowledgements p. 2 Abstract p. 3 Table of Contents p. 4 1. Introduction p. 5 1.1 Vocabulary Learning p. 5
1.2 Psychological conditions leading to word acquisition p. 6
1.2.1 Measuring vocabulary knowledge p. 7
1.3 Vocabulary knowledge and reading in L1 p. 9
1.4 Incidental vocabulary learning in L2 p. 10
1.5 Task-induced involvement load in incidental word learning through reading p. 12
1.5.1 Previous research on the tenacity of the Involvement Load Hypothesis p. 13
1.6 Word exposure frequency in incidental word learning through reading p. 15
1.7 The effects of task involvement load and word exposure frequency explored in one study p. 18
1.8 Aims, research questions and hypotheses p. 19
2. Methodology p. 22
2.1 Participants p. 22
2.2 Materials p.22
2.2.1 Pre-test p. 22
2.2.2 Reading texts p. 24
2.2.3 Operationalization of task-induced involvement load p. 25
2.2.4 Operationalization of word exposure frequency p. 26
2.3 Procedures p. 26
2.3.1 Data collection p. 26
2.4 Assessment: the post-tests p. 28
2.4.1 Time of measurement p. 28
2.4.2 Test purpose and format p. 28
2.4.3 Comprehension questions p. 30 2.5 Statistical analyses p. 30 2.5.1 Research question 1 p. 31 2.5.2 Research question 2 p. 31 3. Results p. 32 3.1 Research question 1 p. 32
3.1.1 Initial word learning p. 32
3.1.2 Medium-term word retention p. 35
3.2 Research question 2 p. 38
4. Discussion and conclusions p. 40
5. Considerations for future research and implications for pedagogy p. 47
References p. 49
Appendices
1. Examples of CATSS items p. 53
2. Pre-test corpus articles p. 53
3. Pre-test item sample p. 54
4. Examples of each task p. 55
5 1. Introduction
1.1 Vocabular y learning
Vocabulary knowledge is an essential component of language since it facilitates the process of conveying a message. Limitations in second language (L2) lexicon are therefore perceived as a serious obstacle in attaining a higher level of proficiency by many learners. Consequently, one of the biggest challenges that L2 learners face is the vast number of words they ought to learn in order to have a high command of their L2. Teachers, on the other hand, face the challenge of how to best facilitate this complex process. When one sets to learn (or teach) a L2, it is desirable to draw learning goals. Traditionally, the two main learning goals in L2 vocabulary acquisition1 concern the vocabulary size and
‘depth’ of vocabulary knowledge. In order to fulfil the first goal, i.e., to understand written or spoken discourse in L2, the learner must know certain percentage of words in the perceived textual or aural input, or she has to have command of an estimated number of word families (i.e., word stems plus all closely related affixed forms) in L2. Recent research has shown that the coverage to comprehend an authentic text in English is 98-99%, which roughly corresponds to a receptive knowledge of 8,000-9,000 word families (Hu & Nation, 2000). Similar research in Dutch revealed that 10,000 headwords are required to understand first-year academic texts (Hazenberg & Hulstijn, 1996). The number of word families that it takes to manage a daily conversation in English amounts to 6,000 – 7,000 (Nation, 2006). Although to successfully function in a L2, large numbers of lexical items are needed, research has shown that the learner has to know a great deal more about these items in order use them properly, hence the second learning goal mentioned earlier. Nation (2001:27) has produced an elaborate account of ‘knowing a word’ (see Table 1). This approach to the depth of word knowledge could be seen as a ‘sum of interrelated subknowledges’ (Laufer & Goldstein, 2004: 400) which firstly stipulates that one has to acquire both form and meaning of a word as well as a proper use of each lexical unit in order to achieve a good command in L2.
The second important aspect of Nation’s description is a clear dichotomy in receptive and productive knowledge type. From an ‘on-continuum’ angle (p. 25), receptive type of knowledge means perceiving a word while listening or reading and retrieving its meaning from the mental lexicon. The term ‘receptive’ is not fully fitting though, because while listening or reading, a concept has to be linked to phonological and/or orthographic features of the word to activate meaning. Productive type of knowledge conveys expressing meaning through speaking or writing and retrieving and producing the appropriate spoken and written word form. An alternative view upon word knowledge defines the two types as different kinds of associational knowledge (Meara, 1990), i.e., passive and active vocabulary.
1Although the terms “learning” and “acquisition” sometimes describe the contrast of learning a foreign vs. acquiring a
6 The former includes items that can merely be activated by external stimuli (i.e.: word forms listened to or read). The latter can be activated by other words, since it has many incoming and outgoing links with them. A point of criticism in this view may be that associations do not solely drive our use of language, but meaning does as well, therefore an object can be actively named in L2 just because one sees it.
Table 1. What is involved in knowing a word (Nation, 2001:27)
Form Spoken Receptive* What does the word sound like?
Productive How is the word pronounced?
Written Receptive What does the word look like?
Productive How is the word written and spelled? Word parts Receptive What parts are recognizable in this word?
Productive What word parts are needed to express this meaning?
Meaning Form & meaning Receptive What meaning does this word form signal?
Productive What word form can be used to express this meaning? Concept & referents Receptive What is included in the concept?
Productive What items can the concept refer to?
Associations Receptive What other words does this make us think of? Productive What other words can we use instead of this one?
Use Grammatical
functions
Receptive In what patterns does this word occur? Productive In what patterns must we use this word?
Collocations Receptive What words or types of words occur with this one? Productive What words or types of words must we use with this
one? Constraints on use
(register, frequency…)
Receptive Where, when and how often would we expect to meet this word?
Productive Where, when and how often can we use this word?
Note: *Receptive and Productive refer to two types of knowledge.
The last important contribution of Nation’s description is its implication for word learning and pedagogy. Whereas word form and meaning might be better learned through intentional learning, collocations and constraints on frequency might be better acquired through extensive reading, thus through a combination of explicit instruction and incidental vocabulary learning (Schmitt, 2008), which will be elaborated on in section 1.4. It is also important to state that the quality (or depth) of word knowledge is affected by cognitive processes in vocabulary learning. Section 1.5 will be devoted to this topic.
1.2 Psychological conditions leading to word acquisition
It has been stated in the previous section that proficient L2 command relies on high quality of word knowledge. But what are the psychological processes that are required to learn and remember a word?
7 Research has identified three main general processes that facilitate word acquisition: noticing, retrieval, and generative (creative) use (Schmidt, 1990; Baddeley, 1990), which will be described in this section in more detail.
Noticing is giving attention to a word as a useful language item, which may be influenced by
salience of the word in the text or spoken discourse, learners’ previous contact with the word, and their realisation that the word fills a gap in their knowledge of L2 (Brown, 1993). Learners also have to be engaged and interested in the chosen content to be acquired to favour word learning. Noticing occurs when ‘learners give attention to a language item as a part of the language rather as a part of a message’ (Nation, 2001:64), which is referred to as decontextualisation and can arise when a word is, for example, explained in definition, as a synonym or translated in L1. Even though decontextualisation is not an essential element in language learning due to its restrictive, explicit character, it can assist learning.
The second general process that may lead to word acquisition is retrieval. Upon noticing a word in a text or spoken input, this word is subsequently successfully retrieved, if the learner is able to elicit its meaning (and form) from the mental lexicon. Retrieval can be thus either receptive, when one perceives the spoken or written word form and has to retrieve the word’s meaning, or productive, when one desires to communicate the word meaning and has to retrieve its spoken or written form. Each retrieval of a word strengthens the path linking form and meaning, which results in an easier next retrieval, therefore not only isolated word repetition, but repeated opportunity to retrieve the word is crucial for word learning (Baddeley, 1990). This can be facilitated through a number of activities such as learners’ negotiation of the word meaning with each other or with the teacher, or through brief explanations of items when listening to a story.
The third major process leading to word being remembered is generation (creative) use, which takes place when words to which the learner has already been exposed are subsequently perceived or used in a new semantic sense of the word, with new inflection, collocation, or in a different grammatical context. The degree of generation is highest when the learner has to reconceptualise her knowledge about the word. Not only production of literal repetitions of a word, but creative production of words in new contexts results in high learning gains and retention (Hall, 1992).
1.2.1 Measuring vocabulary knowledge
Although vocabulary knowledge tests come in all shapes and sizes, the majority of them strive to measure the form-meaning link may it be directly when learners have to show comprehension of the meaning or produce the target form for given meaning (Laufer & Nation, 1999; Schmitt, Schmitt, &
8 Clapham, 2001), or indirectly when meaning associations are assessed (Read, 1993; Schmitt N. , 1999). Most of these vocabulary researchers and test authors also accept the dichotomy of receptive (passive) and productive (active) knowledge of a word. When passive word knowledge is tested in a L2 class, one can, for example, ask for a word’s L1 equivalent. To test active word knowledge, the learner would have to produce a L2 equivalent of a L1 word. Both of the named examples refer to a recall test. When a learner is asked to select the correct answer from a number of word meaning or word form options, we speak of a recognition test. The fact that no full consensus had been found in the distinction between active and passive knowledge being dichotomous (Meara, 1990) or on a continuum (Melka Teichroew, 1982) contributed to Laufer & Goldstein’s (2004) ‘hierarchical’ approach towards the active-passive knowledge distinction. They defined four degrees of knowledge (i.e.: active recall, passive recall, active recognition, and passive recognition) based on two dichotomous distinctions:
(1) supplying the form for a given meaning versus supplying the meaning for a given form (2) being able to recall versus only being able to recognize (whether form or meaning)
How the two above distinctions gave rise to four degrees of knowledge is captured in Table 2.
Table 2. Degrees of vocabulary knowledge and the sequence of degrees of strength (adapted from Laufer & Goldstein, 2004)
Degree of knowledge Test Task Strength degree
Active Recall Supply the L2 word (L1 prompt) (1) strongest
Passive Recall Supply the L1 translation (L2 prompt) (2) or (3)
Active Recognition Select the L2 word (L1 prompt, 4 L2 options)
(2) or (3)
Passive Recognition Select the L1 translation (L2 prompt, 4 L1 options)
(4) weakest
Note: Active = retrieval of form, Passive = retrieval of meaning.
Laufer and Goldstein’s approach attempts both to create a measure of vocabulary knowledge within which some degrees of knowledge are more advanced, or stronger than others, and to avoid traditional measuring of the learner’s vocabulary size in one modality only – passive recognition. Therefore the measure designed by them is referred to as ‘the strength of vocabulary knowledge’ (Table 2). Active knowledge is seen as more advanced than passive, since L2 learners first acquire words passively, and only gradually start to produce them actively in discourse. Moreover, in order to recall a word in L2 or its equivalent in L1 requires a better memory trace compared to recognition, where one can choose from given options, and thus rely on prompts. If learners can recall a lexical
Sequence of the four strength modalities in CATSS
9 item, they can typically recognize it as well. Consequently, there is a ‘sequence’ in the four strength modalities: active recall has the highest level of strength compared to passive recognition with the lowest level of strength. To test this strength-of-knowledge hypothesis, the authors proposed the ‘Computer Adaptive Test of Size and Strength’ (CATSS), which measures vocabulary size and tests each word in accordance with the suggested hierarchy of knowledge strength. The feature of adaptiveness in this computerized test enables to assess a word in the most advanced modality (i.e.: active recall) first. If the learner scores correctly, the word is not tested further down the hierarchy of strength. Since CATSS is a test of vocabulary size, it uses five word frequency levels from the 2,000 most frequent items up to the Academic Word List level (Coxhead, 2000; see more details in section 2.2.1) containing 30 items on each level, which adds up to 600 items on the entire test. The test is available in a bilingual and monolingual version. Test item examples of level 3,000 of the English monolingual CATSS version are given in Appendix 1. The results of Laufer and Goldstein’s study yielded strong support to the hypothesis that lexical knowledge is constructed as a hierarchy of four degrees of strength, with the lowest scores achieved in active recall conform to the premise that it is the most difficult degree of word knowledge to retrieve. The results also provided support to the sequence of the four modalities, which were present in all frequency levels. An additional finding related to the academic success in L2 performance yielded correlations of all strength modalities with the learners’ English class grade, among which the highest correlation was achieved for passive recall (retrieving meaning for a given word).
1.3 Vocabular y knowledge and reading in L1
In the further sections special attention will be given to incidental vocabulary learning in L2. Since lexical knowledge plays a crucial role in the language development and reading skills in general (so not only in L2), it is desirable explore the relationship between word knowledge and language development in the first language.
Before children start to read they acquire words through speech. In general, 1-year-old children gradually develop their holistic and undifferentiated lexical representations (e.g.: Walley, 1993) to more specified ones between their 6th and 8th year of age (Metsala & Walley, 1998). As
children become literate, opportunities for word learning are boosted by their exposure to written language. Research has shown that the amount of reading is the prime contributor to individual differences in children’s vocabularies (Hayes, 1988). Learning to read involves the acquisition of mappings between phonemes and graphemes of words. So words are decoded when one is aware of the distinct speech sounds in spoken words and the systematic relations of written letters to these sounds (Verhoeven & Perfetti, 2011).
10 According to Perfetti’s Lexical Quality Hypothesis (2002; 2007), individuals’ differences in the quality of word representations have consequences for reading and comprehension skills. ‘Lexical quality’ (LQ) refers to the extent to which the reader’s knowledge of a given word represents the word’s form and meaning constituents and knowledge of word use that combines meaning with pragmatic features’(2007:359). Consequently, vocabulary of any language consists of words widely varying in their LQ, from low-frequent words never encountered to high-frequent, commonly encountered ones. LQ thus not only includes the reader’s vocabulary size, but the quality of word representations as well (i.e.: how stable is the reader’s knowledge about the word’s form and meaning). High lexical quality contains well-specified and partly redundant representations of form (orthography and phonology) and flexible representations of meaning, which enables rapid and reliable meaning retrieval. Low-quality representations lead to word-related problems in comprehension (Perfetti, 2007).
As pointed out earlier, to be able to read one must be able to decode word meaning. The development of word decoding can also be fostered by reading. Then it can be seen as an ‘incremental acquisition process’ (Verhoeven & Perfetti, 2011:3), in which the learner gradually, through reading experience, builds up word-specific representations. So the association between vocabulary knowledge and reading skills seems to be reciprocal: vocabulary knowledge enables comprehension, and reading together with comprehension enable word decoding (see study: Verhoeven, van Leeuwe, & Vermeer, 2011).
1.4 Incidental vocabulary learning in L2
When learning a L2, the learner is engaged in the acquisition of sounds, words, structures, and discourse through conscious attention to these units, but she may also acquire them subconsciously (Van Polen, 2013a). Alternatively, and especially in the classroom setting, the conscious, deliberate type of learning usually prevails. Drawing upon this distinction in L2 acquisition, two views have been formed. First, incidental learning that is most often defined as “picking up” words and structures when a L2 learner engages in a communication task such as listening or reading, and when she focuses on the meaning of the language. Secondly, intentional learning, referred to as “the deliberate committing to memory of thousands of words (their meaning, sound, and spelling) and dozens of grammar rules” (Hulstijn, 2003:349). The terms have received various interpretations from researchers (Ellis, 1994b; Huckin & Coady, 1999; Gass, 1999), which are also sometimes identical with the two more general terms of implicit and explicit learning indicating the degree of involvement of consciousness (or awareness) in the act of learning. Although all mentioned terms overlap, they refer to different constructs and different areas of research.
11 Research aimed towards instructed vocabulary acquisition has argued that intentional learning focusing on word form is more profitable than incidental learning focusing on word meaning (De la Fuente, 2006; Laufer B., 2005) resulting in a claim that explicit teaching of contextualized vocabulary is superior to word acquisition that emerges as a by-product of L2 listening and reading (Barcroft, 2009). Moreover, the traditional classroom environment does not seem to facilitate the time needed to acquire language solely incidentally, without focus on its lexical items as parts of a greater system consisting of morphology, syntax, and pragmatics. Although some research of incidental word learning was devoted to learning through listening (Barcroft & Sommers, 2005), the bulk of it focused on learning through reading. The reason is quite logical, since, as already explained in section 1.3, reading is one of the prime contributors to learner’s vocabulary development. Furthermore, it has been established that to comprehend authentic written texts in L2, a learner has to be exposed to massive amount of textual input (Pigada & Schmitt, 2006; Horst, 2005). Contrary to older research reporting on low pick-up rates, Pigada and Schmitt achieved 65% word knowledge enhancement (133 target words tested on spelling, meaning and grammar features) after a month of extensive reading in French. Horst’s investigation indicated that through extensive reading of self-chosen books learners gained new knowledge of more than half of the 100 words that had been rated as unfamiliar by the learners prior to the experiment. Still, the prevailing opinion of scientists in this field is that learning words solely through reading is a rather time-consuming process, prone to errors (Peters et al., 2009).
In his review article on instructed vocabulary learning Schmitt (2008)defined several important issues in incidental vocabulary learning through reading (IVLR), among which are the effectiveness of IVLR, the number of exposures to a word in text, extensive reading, inferencing from context, glossing, and the addition of explicit activities to implicit learning situations. The author concludes that a vast amount of written input is indeed needed to learn words incidentally when reading, that IVLR facilitates better orthographic and grammatical knowledge of partially known words, and that due to higher learning and retention rates in recognition rather than in recall IVLR stimulates the development of ‘partial rather than full level of mastery’ in L2 vocabulary learning (p. 347). In his proposal for successful vocabulary instruction, Schmitt therefore suggests a combination of intentional and incidental approach that ‘positively require each other’ (p. 353). To acquire all levels of word knowledge including its collocation use and constraints of register and frequency is impossible without a large exposure to listening and reading of authentic texts. On the other hand, words learned incidentally might not lead to productive word knowledge without additional attention to intentional learning.
It can be derived from the previous sections of this thesis that cognitive processes involved in the acquisition of receptive word knowledge needed for reading (or listening) differ from those needed to speak or write, thus to develop one’s productive word knowledge. A research strand that investigates
12 the aspects of receptive and productive vocabulary acquisition in L2 usually uses written input that manipulates certain text features (e.g.: highlighting, glossing, lists of words to notice) and sets conditions that deepen learners’ mental involvement with unfamiliar words by means of additional (post-) reading tasks such as gap filling, matching words with synonyms or definitions, multiple-choice or open comprehension questions, and composition writing (Hulstijn & Laufer, 2001; Peters et al., 2009). Moreover, it has also been suggested above that learners learn and retain words well when exposed to them repeatedly in authentic texts, and when this repeated exposure is strengthened by giving the learner plentiful opportunities to retrieve the lexical item (Baddeley, 1990; Nation, 2001). It can be therefore ascertained that word exposure frequency is another crucial factor promoting IVLR. In the following sections each of the two factors essential for the acquisition of vocabulary when reading, task-induced involvement and word exposure frequency, will be elaborated on in more detail
1.5 Task-induced involvement load in incidental word lear ning through reading
In the attempt to shed light on the cognition of vocabulary acquisition, Craik & Lockhart (1972) proposed the concept of depth of processing stipulating that the storage of a piece of information in long-term memory is determined by shallowness or depth with which it is initially processed, and not by the length of time that it is held in short-term memory. They claimed that there are several levels of processing: lexical items are processed at a deep level, but the processing of phonological forms occurs at a shallow level.
The depth-of-processing hypothesis was later acknowledged by Hulstijn & Laufer (2001) who, however, sought the operationalization of this cognitive term in L2 vocabulary learning tasks. Their
Involvement Load Hypothesis (ILH) proposes a motivational-cognitive construct of involvement load
that consists of three components (need, search, and evaluation) with each having two degrees of prominence (moderate and strong). ‘Need’ is the motivational dimension of involvement. It is moderate when imposed by an external agent (teacher asks to use a word in a sentence), it is strong when intrinsically motivated (learner’s decision to look up a word in a dictionary). The first cognitive component, ‘search’, reflects the attempt to find the meaning of an unknown L2 word, or to find the L2 word form expressing a concept. The second cognitive component is ‘evaluation’ that comprises a comparison of a given word with other words, or a meaning of a word with other meanings of the word. The combination of the above components and their degrees of prominence makes up involvement load expressed in terms of involvement index (moderate = 1, and strong = 2). Not every task demands all three components to be utilized by the learner. A task with a strong degree of prominence of these elements has a higher level of involvement load. Drawing upon the ILH, high
13 learning gains should be attributed to the amount of mental effort required by the learner, exploiting a deeper level of processing.
1.5.1 Previous research on the tenacity of the Involvement Load Hypothesis
In this section the results of three studies will be discussed to elucidate the tenacity of the ILH.
Hulstijn and Laufer (2001)
The aim of the study was to test short and long-term retention of ten words through incidental learning based on the degree of task-induced involvement load. The subjects were advanced English foreign language (EFL) learners in the Netherlands and Israel, divided in three groups in each country, every group was assigned to one task. Three tasks were designed to test the hypothesis. Task 1 consisted of reading a text equipped with marginal L1 glosses to target words. The students were asked to answer multiple-choice comprehension questions after having read the text. Task 2 involved reading plus gap filling of target words followed by comprehension questions. Task 3 required the subjects to write a composition using all ten target words. The involvement index comprised a sum of the need,
search and evaluation component values. Task 1 induced moderate need (value: 1), but no search or
evaluation, so the involvement index of this task was 1. Task 2 elicited moderate need (value: 1), no search, and moderate evaluation (value: 1). The involvement index of this task was 2. Task 3 required moderate need (value:1), no search, and strong evaluation (value: 2). Therefore the involvement index of this task had the highest value (3), due to the requirement to use the target words in an original context.
The results of the experiment fully supported the ILH only with regard to the native Hebrew group. Task 3 induced the highest level of involvement load and it yielded the best immediate and delayed retention rates. The gains in the Dutch-English experiment were the highest in the composition group (Task 3), but no significant difference was found between the mean scores of the reading-only and reading-plus-fill-in groups (Tasks 1 and 2). The composition task put increased demands on the learners’ linguistic resources, which follows the prediction that learner’s higher mental involvement in learning a word induced by the task will provide better retention. The ILH does not, however, predict whether an output task will lead to better gains than an input task. All in all, the ILH tested in this experiment did not provide conclusive findings for the generalizability of the construct to all tasks at all language levels.
Keating (2008)
14 (Folse, 2006among the few studies). Moreover, the research suffered from some limitations such as predominant focus on advanced learners, experiments testing the task-induced involvement in receptive, but not productive, knowledge, and disregard of a time-on-task criterion in the gain scores. Keating (2008) recognized the three drawbacks and implemented them in the main research questions of his study. Undergraduate beginner learners of Spanish were tested on two tasks identical to those in Hulstijn and Laufer’s research, and Task 3 involved original sentence writing with the use of target words without prior text reading. Immediate and delayed post-tests of passive and active knowledge2
were administered, the scores of which were converted to reflect words learned per minute.
The findings showed that low proficient learners benefited the most from tasks requiring higher involvement load (i.e., gap filling/Task 2 and sentence writing/Task 3). Higher retention scores of passive knowledge were obtained in Tasks 2 and 3 compared to Task 1 (with low involvement load). These two findings lent support to the ILH. Active knowledge gains on the immediate post-tests confirmed the predictions of the hypothesis, which, however, the results of delayed post-tests did not yield for Task 3. The last RQ concerning the task completion time could not be answered positively either since the benefits of high involvement load tasks ‘did not hold when test scores were adjusted to reflect time on task.’(p. 380) On the whole, the study extended the empirical support for the ILH as far as low-proficient learners and their benefit from more involving tasks are concerned, but it did not support the prediction that tasks yielding high vocabulary gains lead to more words learned per minute.
Kim (2011)
Kim attempted to prove the ILH by exploring the correlation of language proficiency levels, learner’s cognitive involvement and retention of target words. Two experiments were carried out with L2 learners of English at two proficiency levels3. Experiment 1 tested the impact of involvement load
across two proficiency levels in a reading task, a gap-fill task, and a composition task. Experiment 2 investigated the effect of two different post-reading productive tasks (writing composition and writing sentences) with the same level of involvement load4 on the initial learning and retention of target
words by L2 learners with different proficiency levels.
The results of Experiment 1 supported the ILH partially. Learners’ gains were greatest in tasks requiring a higher level of involvement load (composition), but no significant difference was established between the scores of the reading and gap-fill groups. A strong degree of prominence of
2Passive recall post-test was administered in the form of Spanish – English word translation task, and active recall as
English-Spanish sentence translation task.
3 The subjects were ESL students of a US undergraduate degree program and an Intensive English Program (IEP) followed
prior to the enrolment in a degree program. The proficiency levels corresponded to these two groups based on their TOEFL scores: Undergraduate above 520 (upper intermediate/advanced), IEP between 470-520 (intermediate/ upper
intermediate).
4 According to ILH this level was established as identical for both composition tasks, however, Kim intended to address the
15 the evaluation component of ILH was proven to be the most influential factor for initial vocabulary acquisition. Based on the data of Experiment 2, both tasks had the same effect upon initial learning as well as retention, which lent support to the ILH. Even though composition writing might require deeper cognitive processing than sentence writing, the results show that both self-generated tasks use the same level of involvement load. The effect of proficiency investigated in both experiments did not produce significant difference on task between the two different proficiency groups. ‘This finding provides some support for Hulstijn & Laufer’s claim that less proficient learners can also perform an involving task.’ (p. 126) It is important to add though that these ‘less proficient learners’ were not beginners in the proper sense of the word. Although Kim’s study replicates some findings of Hulstijn and Laufer’s experiment, it cannot be concluded that it provides grounds for a broad pedagogical application, which is due to the limited number of participants (N=64) and tasks (3) that were only performed once by each participant. As far as the validity of the ILH is concerned, Kim expresses his doubts towards the involvement index established as a sum of the three ILH components: need, search and evaluation. It is a fact that separate tasks can occur, in which the involvement index is the same, but the distribution of the components is different. The question then arises whether the hypothesis can be supported for both tasks and whether equal learning gains will be guaranteed.
Despite the fact that there is more indirect than direct evidence of ILH, the common finding of cited research is that post-reading vocabulary-focused tasks with high involvement load result in high learning gains. Composition or original sentence writing tasks demand the learner to make decisions on how additional words will combine with the new word in the output, which involves high degree of mental effort, hence a strong evaluation component of involvement.
1.6 Word exposure frequency in incidental word learning through reading
Incidental word learning is a cumulative process that requires repeated encounters with (unfamiliar) words in a single or multiple texts (Paribakht & Wesche, 1999; Nation, 2001). In the course of the last 15 years a number of studies have been conducted exploring the effect of word exposure frequency on incidental vocabulary learning through reading. Some of the findings will be discussed in this section.
Waring & Takaki (2003)
The study tested Japanese lower-intermediate learners of English and their immediate learning gains and delayed retention of words that occurred with the following frequencies in a text: once, 4-5 times, 8 – 10, 13-14, and 15-18 times. The reading material was a 400-word graded reader with a
16 “known words” coverage of 96.2% of the running words (following the broader criterion of optimal rate between 96 – 99%, (Laufer, 1989; Hu and Nation, 2000). In order to preserve a decent spread of the word occurrence, to ensure that the participants did not know the target words (TW) prior to testing, and that they would not encounter them between the experiment and the delayed post-test,
substitute words were inserted in the reader that referred to the same concepts as the real TWs, but
had their spelling changed (e.g.: yes – yoot, sun – blund). The vocabulary acquisition assessment was conducted by three tests: recognition of form / active recognition (select L2 words you met in the text), recognition of meaning / passive recognition (select nearest meaning to L2 target from four L2 options), and recall of meaning / passive recall (submit L1 translation for L2 prompt).
The immediate test scores were highest in the active recognition test (61%). In a delayed post-test administered three months later, active recognition score was still the highest, but diminished drastically (33.6%). None of the items encountered less than 8 times in a text was retained after three months on the recall test. The authors conclude that learners have to meet a word at least eight times to have a 50% chance to recognize its form three months later. To be able to score 15% correct on a passive recall test of the same word, one has to be exposed to it more than 18 times in textual input. ‘The benefits of the learning were soon lost especially when higher demands were placed on word knowledge (i.e., when meaning was not prompted)’ (p. 149). Among other possible factors, the reasons for the low pick-up rate were explained by the learners’ focus on comprehension instead on words, and by the difficulty to learn substitute words because they were synonyms to real English words already known to the learners. The study revealed some limitations such as the relatively low number of subjects (n = 15), limitations of the reading material in terms of its variety and length to gather sufficient data on the effectiveness of IVLR, and the fact that the used graded reader lacked optimal spread of words of various occurrence rates, whose meaning is equally easy to be inferred from context.
Pigada and Schmitt (2006)
One limitation of previously discussed study was addresses in the research by Pigada and Schmitt three years later. Their 1-month case study investigated a Greek learner of French reading four pocket-size books, which equalled to a corpus size of approximately 30,000 words. The acquisition of three aspects of word knowledge was investigated: spelling, grammatical behaviour, and meaning. The 133 TWs occurred in the following frequencies in the texts: 1, 2-3, 4-5, 6-10, 10+, and 20+ times. The test method was one-on-one interview consisting of a spelling and meaning-and-grammar test allowing to measure partial knowledge on these aspects of word knowledge as well (i.e., marking scheme: 0 – 1 -2).
The results showed relatively high learning gains in spelling (52.3%) and grammar (27.6%), but the learning of meaning after a month achieved the lowest score (20.2%) out of all word knowledge
17 aspects. In terms of word exposure frequency and word meaning learning, ‘by about 10+ exposures, there does seem to be a discernable rise in the learning rate (36.3%)’ (p. 18). By 20+ encounters uptake rates of 60% were achieved by some words. However, the meaning of some words was not acquired even after the learner met them 20+ times in the text. Two possible explanations were offered for this result. Firstly, the learner already had partial knowledge of some high-occurrence words, which could have resulted in less room for improvement (i.e., a ceiling effect). Secondly, some of the words occurring only 1 to 10 times in the texts were easier to learn due to their morphology, salience, or the context in which they were encountered. It is important to emphasize that although the study showed greater gains than previous studies in vocabulary learning through extensive reading, the majority of the achieved gains were related to spelling and grammatical categories of nouns and verbs, not to meaning. An important contribution of the study was the use of various measurement technologies ‘to illustrate a degree of vocabulary acquisition previously undemonstrated’. (p. 21).
Rott (2007)
The study researched short-term learning gains and long-term word retention of 24 TWs that (38) English university students learning German were exposed to once or four times in three medium-length texts. The second independent variable was designed in the form of visual enhancements resulting in three reading conditions: TWs glossed four times (4G), TWs first glossed then retrieved in L1, and bolded twice (GR), and TWs first glossed and then bolded three times (GB). Among several hypotheses that were formulated the one dealing with frequency of word encounters predicted higher learning gains in receptive and productive word knowledge for words that occurred four times in the text. An adapted paper-and-pencil version of the CATSS (see section 1.2.1 for explanations) was used to measure active and passive recall and passive recognition.
Results revealed that when readers encountered the TWs four times in the GR and 4G condition, they gained significantly more receptive and productive word knowledge than when they met a TW only glossed once in the reading passage. These results were obtained on both immediate and delayed post-tests. So the findings only partially supported the above hypothesis: visual enhancement (i.e., 4 times bolded TWs) did not significantly influence word learning compared to semantic enhancement in the form of additional glossing (4 times glossed) and word retrieval (4 times glossed, retrieved in L1, bolded twice). Besides the effect of exposure frequency, this study yielded positive support for techniques enhancing meaning in the text. A gloss as well as gloss followed by the task to translate a TW to L1 are both ‘obtrusive interventions’ (p. 172) that appear to have strengthened learners’ word encoding. When they look for the gloss in the text margin, processes of isolation and rehearsal are fostered in the learners’ working memory that boost mapping the word’s form to its meaning. Moreover, the L1 translation even reinforces the L2-L1 word association, and so it contributes to long-term retention of word meaning. The author concluded that ‘repeatedly directing
18 learners’ attention to semantic word enhancements within the same text has clear advantages for the development of receptive and productive word knowledge’ (p. 188).
The presented research underscores a widely-held view that vocabulary acquisition is a gradual and incremental process, and that full word knowledge development is dependent upon repeated encounters with unfamiliar or partially known words in one or multiple texts. At the same time, the cognitive prerequisite to learn, remember and use a word is conditioned by establishing, strengthening and retaining form-meaning connections during repeated encounters with the word (Eckerth and Tavakoli, 2012). The investigation into the effects of both factors, mental involvement and word exposure frequency, upon word learning when reading will be the main subject of the next section.
1.7 The effects of task involvement load and word exposure frequency explored in one study
Research discussed so far treated each variable, involvement load and exposure frequency, separately. In order to fill this gap in research, the two factors were combined in one experiment conducted by Eckerth and Tavakoli in 2012, which will be explained in this section. The main aim of the study was to explore the effects of word exposure frequency and task-induced involvement load upon immediate word learning and medium-term word retention in L2 English in the context of incidental vocabulary learning through reading. The study addressed three research questions (RQs):
(1) What are the effects of exposure frequency and task-induced involvement load on learners’ overall initial word learning and medium-term word retention?
(2) What are the effects of exposure frequency and task-induced involvement load on learners’ two types of word knowledge (active recall and active recognition) in initial word learning and in delayed word retention?
(3) Do word retention rates differ for the two types of word knowledge measured?
Data was collected and analysed by means of ‘a within-group design, so that each single participant was exposed to each of the target words in each of the three reading conditions, while time on task was controlled for’ (p. 231). Advanced learners of English (N = 30) with Asian, Slavic, and Romance L1 in a pre-university UK learning programme were exposed to three reading texts (approx. 1,200 words long) in consecutive three weeks. Each text contained 10 TWs (30 TWs in total). These words were established as unknown to the learners prior to the main experimental treatment. To preserve the condition of incidental learning, the participants were not informed about the specific aims of the study, neither was any of the vocabulary post-tests pre-announced. The exposure frequency was operationalized on two levels: one and five occurrences of TWs in every text. The task-induced
19 involvement load was operationalized by three reading tasks differentiated by the depth of involvement from low to strong as follows: reading + glosses (Task 1, index = 1/low), reading + gap filling (Task 2, index = 2/moderate), and reading + composition writing (Task 3, index = 3/strong). Word learning gains and retention were assessed by the monolingual version of CATSS (Laufer et al., 2004, discussed earlier in section 1.2.1) in all four degrees (types) of vocabulary knowledge: active and passive recall, and active and passive recognition. These post-tests were administered immediately after each reading session and three weeks after the last treatment.
The results revealed that, firstly, higher frequency of words and higher involvement load resulted in higher overall immediate learning gains. The delayed post-test showed that scores on Task 3 were significantly higher than scores on Tasks 1 and 2, however no significant effects were found for word frequency (RQ1). In other words, over time the effect of frequency ‘tends to fade, whereas the effect of task-induced involvement load is more time-stable.’ (p. 242) Secondly, overall word gain scores in both post-tests were highest on active recognition and lowest on active recall. Involvement load did not alter this overall score distribution in the immediate post-test, but affected the distribution in the delayed post-test, where Task 3 yielded higher active and lower passive word retention than the other two tasks. On the other hand, word exposure frequency affected the immediate test scores, but this effect ceased in the delayed post-test, so ‘the effects of frequency were less time stable’. (p. 242; RQ2). Thirdly, all four knowledge types decreased significantly over time. Word knowledge type and time showed significant interaction. In terms of absolute word gains on the immediate and delayed post-tests, the recognition type of knowledge was retained better over time. However, the results can be read alternatively in terms of how great was the loss of knowledge over time. Since the overall scores in recognition type of knowledge (initial and medium-term) were significantly higher than those in recall, this should suggest that where ‘more was gained initially, the more was lost subsequently’. (p. 243) Therefore, in terms of knowledge loss, recall was retained better than recognition over time (RQ3).
To sum up, the results of the study suggest that both variables significantly affect incidental vocabulary learning through reading, but that as time progresses the cognitive involvement based on elaboration of word processing (i.e., an input-and-output task) has a stronger impact than does word exposure frequency.
1.8 Aims, resear ch questions and hypotheses
The research to be presented in this thesis is a replication study of the investigation conducted by Eckerth and Tavakoli (2012), which will be further referred to as the original study. Before the
20 methodology, results, statistical analysis and conclusions will be treated in the following chapters, the aim, research questions and hypotheses of the current study will be presented. The main aim of the present study was identical to the original study: to investigate the effects of task-induced involvement load and word exposure frequency on incidental vocabulary learning through reading in L2.
The current study addressed two research questions:
1. What are the effects of exposure frequency and task-induced involvement load on learners’ two types of word knowledge (active recall and active recognition) in initial word learning and in delayed word retention?
(repeating RQ1 and RQ2 in Eckerth & Tavakoli, 2012)
2. Do word retention rates differ for the two types of word knowledge measured?
(repeating RQ3 in Eckerth & Tavakoli, 2012)
To preserve the incidental mode of learning, the main goals of the study were not communicated to the students, neither were they told that they would be tested on newly acquired vocabulary in the coming weeks. The within-group design that was congruent with the original study allowed the researcher to select TWs as a random factor, not influenced by the characteristics of the reading texts, as the participants were exposed to sets of the same TWs under all three reading conditions, but in three various texts (see section 2.3.1). Time on task was regarded as ‘inherent property of the task’ (Hulstijn and Laufer, 2001:549), so it was not regarded as a separate variable.
Due to factors that will be closer explained in chapter 2, there are several differences in the design of the currently presented research. The participants were Dutch upper-intermediate learners of English. The reading texts adopted in the main experimental treatment were shorter consisting of 700 to 950 words, which resulted in two exposure frequency conditions: 1 and 4 occurrences that we felt were appropriate to the medium text length. The fact that the digital version of the CATSS (Laufer et al., 2004) was not available to the researcher, neither had she enough time and resources to modify it for the purposes of the study (e.g.: using specific Business English vocabulary), led to the employment of a modified paper-and-pencil version of the monolingual CATSS, in which merely two degrees of word knowledge were assessed: active recall and active recognition (see section 2.4.2 for explanations). The last difference to the original design is the time elapsed between the last immediate post-test and the delayed post-test, which was four weeks.
21 The following hypotheses were formulated:
1.a Better overall test scores in active recognition will be achieved compared to active recall in both initial learning and medium-term retention conditions.
1.b Strong level of involvement load (Task 3) will lead to highest scores in both immediate and delayed post-tests for both active knowledge types.
1.c High word exposure frequency (words with 4 occurrences) will result in better immediate post-test scores for both knowledge types: active recall and active recognition, but will not affect the scores in the delayed post-test.
2. Active recall knowledge will be retained better over time than active recognition, and there will be a greater loss of knowledge in recognition compared to recall.
These hypotheses were based on the findings of the original study presented in section 1.7 as well as on the theoretical premises and previous research regarding the strength of word knowledge (section 1.2.1), task involvement load (section 1.5) and word exposure frequency (section 1.6).
22 2. Methodology
To address RQ1 and RQ2 introduced in chapter 1, a within-group design was chosen containing the following independent variables: task-induced involvement load reflecting the depth of processing (Craik and Lockhart, 1972; see section 1.5), word exposure frequency referring to the number of occurrences of a TW in a reading article (see section 1.6), word knowledge type measured on two active degrees of knowledge: recall and recognition (see section 1.2.1), and time of testing divided in immediate and delayed post-tests. The operationalization of these variables will be disclosed in sections 2.2.3, 2.2.4, 2.3 and 2.4 respectively.
2.1 Participants
The participants in the current study were 51 Dutch undergraduate students following a compulsory English 3 course (further referred to as EN3) in Business English5 within their second year of study of
the BSc programme in Business Administration at the University of Applied Sciences in Amsterdam. The participants learned English as a foreign language through instruction at school and they had not lived in an English-speaking country. Although the participants learned English as a foreign language (EFL) outside the context of an English-speaking country, the commonly used term of English as a second language (L2) has been used here. None of the participants was a native-speaker of English and their age of acquisition was at around 10 years of age due to the Dutch education system. The participants’ L2 proficiency level was approximately upper-intermediate, which is equivalent to B2 level6 of the Common European Framework of Reference (CEFR) based on the students’ test scores
achieved in the Oxford Placement Test they had passed as a compulsory part of the programme. It should be stated, however, that it was not feasible to guarantee by the placement test, which only assesses receptive vocabulary and grammar knowledge, that all students fulfilled the B2 proficiency level in both receptive and productive language skills.
2.2 Materials 2.2.1 Pre-test
In accordance with the original study, a prerequisite of the presented study was to ensure that TWs were (nearly) unknown to the students prior to experiment, since the aim was to measure how many new words students learned through reading and how many they would remember four weeks after
5Business English is a variant of English for Specific Purposes that is composed of ‘a mix of specific content (relating to a
particular job area or industry), and general content (relating to general ability to communicate more effectively, albeit in business situations).’ (Ellis & Johnson, 1994)
6B2 Independent User/ Reading: S/he can understand longer texts of middle difficulty using restricted general and
23 the last reading session. This was accomplished by three pre-tests that all groups took prior to the main task.Since the first pre-test only yielded 14 out of the desired 30 TWs that were unknown to at least 70%7 of the participants, two follow-up pre-tests using new test items were taken in weeks 5 and 6.
These tests were a prerequisite for the choice and manipulation of the target words in the reading texts.
a. Purpose and format
The purpose of the pre-test was to establish a pool of words in general and Business English that at least 70% of the subjects had no receptive knowledge of at the time of testing. In other words, the participants should not recognize these words when exposed to their orthographic form, so their semantic and morphological knowledge of the word should be weak or non-existent (Van Polen, 2013b). The pre-test was constructed as a multiple-choice (MC) word recognition test measuring written receptive vocabulary. The stimulus provided the word form, and the test-taker had to access the meaning available in his/her mental lexicon matching the stimulus in order to answer the item correctly. The MC format based on the Vocabulary Size Test (Nation & Beglar, 2007) offered four answer options for each test item that had equal level of word frequency (further explanation is given in paragraph Stimuli). This helped to control the level of difficulty of the answers tapping the same degree of word knowledge, and to make learners demonstrate knowledge of each item.
b. The corpus
Vocabulary corpus (8,114 words) to develop the pre-tests was built on 11 articles dealing with three topics related to the EN3 course: Microfinance, Late Payers and Takeovers. For the complete list of the articles see Appendix 2. In accordance with the content of the EN3 course the subjects were following at the time of the experiment, the corpus consisted of authentic articles from Business English course books, on-line business periodicals, case studies and business theory books.
c. Stimuli
From the corpus mentioned above, a list of 78 test words was generated. The second and third pre-tests contained a narrower selection of 26 test items from two articles of the corpus. All items were selected by range, i.e., a word had to occur in as many texts as possible, and by frequency. Words were ranked from high to low frequency of occurrence checked against Academic Word List (AWL; Coxhead, 2000) and British National Corpus (BNC; 2007). AWL is a specialized vocabulary list of 570 word families outside the most frequent 2,000 words of English (West, 1953), which appear reasonably frequently in a wide range of academic textsregardless of the subject area, and which has a 10.0% total words (token) coverage of the Academic Corpus (Coxhead,1998). The AWL words are grouped into ten Sublists that reflect word range and frequency. RANGE program developed by Paul Nation (2005) was
7This intended ‘non-knowledge’ rate is lower than reported in the original study (Eckerth & Tavakoli, 2012), where only
24 used to obtain lists of words ranked by word frequency and range. Each pre-test contained single-word stimuli and four answer options (see examples in Appendix 3). Every stimulus was presented in a semantically non-explanatory context and indicated part of speech (e.g.: ‘The payment is voluntary’ indicates an adjective stimulus). In this way the learner attention was drawn to the word form and to merely one meaning of the word (limiting homographs or very different meanings of a word). The four answer options were formally insertable in the stimulus sentence. The words used in them were of higher frequency than the stimulus word (i.e., the first two 1,000 word levels of General Service List; West, 1953). Correct answers should therefore reflect the fact that the learner had ‘a moderately developed idea of the meaning of the word’ (Nation & Beglar, 2007:11).
The choice of stimuli was controlled for lexical frequency. One third of the words were highly-frequent, two thirds were low-frequent according to AWL and BNC. To check whether the selected words were equal to or beyond the students’ proficiency level, one of the world’s largest databases of authentic written and spoken language, Cambridge English Vocabulary Profile (2012) was consulted. The ratio 1:2 signifying highly frequent : low frequent words was opted for since it was desirable to obtain a pool of words learners had no receptive knowledge of, thus being low frequent. Inclusion of more common words (i.e., highly frequent) should, on the other hand, encouraged the less motivated learners to complete the test. The tested words were also controlled for word origin, since Dutch, sharing the Germanic language family with English, contains a large deal of cognates8. In order to make
sure that the students neither knew the words, nor did they guess their meaning due to orthographic similarities with their L1, cognates were excluded from pre-test. The word cognate status was checked by VanDale Groot Woordenboek Engels – Nederlands (Martin et al., 1998)
2.2.2 Reading texts
Three medium-long expository texts of 700 - 950 words were used in the main experiment. For this purpose the texts were selected from the pre-test corpus. They were abridged and equipped with the TWs identified as unknown by the pre-test (for full article references see titles no. 3, 5 and 9 of Appendix 2). As opposed to a decontextualized method of vocabulary testing, these medium-long texts provided the learner with more context to infer meaning and the researcher with more opportunity to manipulate the texts by inserting some of the TWs in the text more than once. The articles fulfilled the criteria of representativeness, organization, and large enough size of a well-designed corpus. Each modified text was checked by the researcher (near native-speaker and English language lecturer) and two experienced English native-speaker lecturers on linguistic difficulty and coherence. The 30 TWs applied in the reading texts are shown in Table 3.
8Cognates are L2 words that share meaning and parts of their orthographic or phonological form with their L1 translation.
25 Table 3. Target words of the three reading texts divided by word exposure frequency
T ext High F re q u e n cy (4 oc cu r re n c e s ) Lo w F r eq u en cy (1 oc cu r re n c e ) Mi cro f in an ce entity venture prevail collateral domestic severe rigid austerity argue promptly Lat e Pa ye r s worsen compliance pursue reluctant reiterate breach par enshrined redress statutory Ch in e s e Ta ke o ve r s solicit spur opaque consent wistfully diffuse arbitrary autocrat deference monolith
2.2.3 Operationalization of task-induced involvement load
The independent variable of task-induced involvement load refers to the amount of mental effort required by the learner to accomplish a (post-) reading task. Following this premise, three reading tasks were designed. Task 1 consisted of reading a text equipped with marginal glosses of 10 TWs. Task 2 was designed as a text containing gaps, in which the TWs had to be filled. A word list of all TWs was provided containing the TW, a reference to its part of speech (noun, verb, adjective, or adverb) and its definition. In all three tasks the word definitions were adapted from Collins Online English Dictionary, (2014). Due to the medium length of the texts, it was not feasible to insert four gaps for the high-frequent TWs. In that case, learners would be deprived of too many cues, resulting in a non-intelligible text. Consequently, the high-frequent words had to be filled in twice instead of four times. Task 3 involved reading with marginal glosses with a post-reading composition writing. Students had a choice between writing a text summary or a short argumentative essay dealing with an aspect of the text. For a successful accomplishment of the writing task it was vital to use all 10 TWs in the composition, which was explicitly stated in the task description. Examples of each task are provided in Appendix 4.
The three tasks differ by the level of involvement. The motivational component of involvement, ‘need’, was moderate by all three tasks, since the need to understand or use a TW is imposed by the task, not by the learner. The first cognitive component, ‘search’, was absent in all three cases, since none of the tasks required the learners to look up the word meanings or their translations in a dictionary. The definitions were provided in glosses or in a word list. Finally, ‘evaluation’, the second cognitive factor of involvement, was absent in Task 1. It increased to a moderate level in Task
