The effects of reading, listening, and reading-while-listening on incidental vocabulary acquisition by L2 learners

The effects of reading, listening, and

reading-while-listening on incidental vocabulary acquisition by L2

learners

Amber Dallinga S2865149

MA thesis, Department of Applied Linguistics, Faculty of Arts,

Rijksuniversiteit Groningen

Supervisors:

prof. dr. Marjolijn Verspoor prof. dr. Marije Michel

15.413 words June 18, 2020

Declaration of Authenticity

MA-thesis Student name: Amber Dallinga

Student number: S2865149

PLAGIARISM is the presentation by a student of an assignment or piece of work which has

in fact been copied in whole, in part, or in paraphrase from another student's work, or from any other source (e.g. published books or periodicals or material from Internet sites), without due acknowledgement in the text.

TEAMWORK: Students are encouraged to work with each other to develop their generic

skills and increase their knowledge and understanding of the curriculum. Such teamwork includes general discussion and sharing of ideas on the curriculum. All written work must however (without specific authorization to the contrary) be done by individual students. Students are neither permitted to copy any part of another student’s work nor permitted to allow their own work to be copied by other students.

DECLARATION

• I declare that all work submitted for assessment of this MA-thesis is my own work and does not involve plagiarism or teamwork other than that authorized in the general terms above or that authorized and documented for any particular piece of work.

Signed: Amber Dallinga

Table of Contents

Abstract ... 4

1. Introduction ... 5

2. Background ... 8

2.1 Incidental vocabulary learning ... 8

2.2 Reading ... 12

2.3 Listening ... 14

2.4 Comparing reading and listening ... 16

2.5 Reading-while-listening ... 18

2.6 Comparing RWL to reading and listening ... 19

2.7 Statement of purpose ... 21 3. Method ... 23 3.1 Participants ... 24 3.2 Materials ... 25 3.3 Procedures ... 28 3.4 Analysis... 30 4. Results ... 31

4.1 Overall vocabulary scores ... 31

4.2 Scores per vocabulary knowledge type ... 32

4.3 Retention ... 39

5. Discussion ... 48

5.1 Incidental vocabulary acquisition ... 48

5.2 Incidental acquisition of different types of vocabulary knowledge ... 51

5.3 Retention of incidentally acquired vocabulary knowledge ... 54

5.4 Practical implications ... 57

5.5 Limitations ... 59

6. Conclusion ... 61

Appendices ... 64

Abstract

The incidental acquisition of new L2 vocabulary can be an important contribution to the process of second language development. Previous studies have found multiple modes of L2 input to be effective in facilitating incidental vocabulary acquisition, but results are more mixed regarding the question of which mode of input is the most advantageous for incidental foreign language learning. The current study investigated the incidental vocabulary learning outcomes of target words encountered by Dutch university students with high L2 English proficiency levels through texts that they read, listened to, or read while simultaneously listening to the text. The results of the immediate post-test showed the listening condition to lead to significantly less vocabulary acquisition than the reading condition and the reading-while-listening (RWL) condition. It was also found that significantly lower form recognition scores were obtained in the listening condition than in the RWL condition. For meaning recognition scores as well as meaning recall scores, significantly lower scores were found in the listening condition than in both the reading and RWL conditions. The effect of input condition was the smallest for form recognition, and the largest for meaning recall, for which a moderate effect was found. On the delayed post-test two weeks later, the highest scores were obtained for words acquired in the RWL condition, and the lowest for the listening condition, and this difference was found to be significant. The form recognition scores on the delayed post-test were found to be significantly higher for RWL than for reading. This

suggests that stronger mental representations of incidentally acquired words are created when they are learned through RWL input, which leads to the best long-term retention of the acquired word knowledge.

1. Introduction

There are many different ways in which a second or foreign language (FL) can be learned. Some learners may assume they need to enroll in a language class or follow an online course in order to improve their L2 proficiency. However, these types of formal and explicit learning are not the only possible manners to acquire language. Many foreign language learners experience what is called incidental learning. Incidental learning can be defined as language acquisition that occurs without the intention of learning (Hulstijn, 2013). Incidental learning is often a by-product of another activity (Restrepo Ramos, 2015). For example, incidental vocabulary learning may take place when a learner is reading a text in their L2, encounters a word that is unknown to them, and is able to infer the meaning of the word from the context (Restrepo Ramos, 2015). The learner may need to encounter the new word multiple times in order to memorize the meaning, but the presence of the word in the input can lead to the learner eventually being able to recognize the word and even use in it their own L2 production. In fact, a study by Ahmad (2012) even suggests that incidental word acquisition leads to better vocabulary test results than intentional vocabulary learning. This could be because intentional learning only entails simply memorizing a new word and its meaning, while incidental learning involves more cognitive processes, such as inferring the meaning of the unknown word in order to comprehend a text (Ahmad, 2012).

Different types of media can provide the FL learner with input that can facilitate second language acquisition (SLA) through incidental learning, such as movies and television shows, music lyrics, and video games. However, the most widely researched and most

commonly used mode of input for incidental acquisition is through written text. Multiple studies have found evidence that reading texts in the L2 can lead to incidental vocabulary gains (Day, Omura, & Hiramatsu, 1991; Restrepo Ramos, 2015; Swanborn & de Glopper, 1999). Similar results have been found in studies investigating incidental vocabulary learning

from listening to texts being read aloud by someone else (Al-Shehri, 2015; Ellis & Le, 2016). In addition to merely reading or listening to a text, it is also possible to do both of these at the same time, and read along while a text is being read out loud.

These differences in the reception of the input may lead to differences in the amount or type of incidental vocabulary learning that can occur. For teaching purposes, as well as for individual learners aiming to expand their L2 vocabulary, it would be useful to know which mode of input is the most facilitative for incidental vocabulary learning, and whether different types of input have different benefits. Most studies report advantages for reading over listening (Chang & Ma, 2018; Hatami, 2017; Vidal, 2011), although no differences in vocabulary gains were found between reading and listening input by Feng and Webb (2019). When comparing reading-while-listening (RWL) to reading, Teng (2018) found higher incidental vocabulary gains for RWL than for reading. This was also found by Webb and Chang (2012) and Brown, Waring, and Donkaewbua (2008). However, Tragant Mestres, Llanes Baró, and Pinyana Garriga (2019) found no differences in incidental vocabulary acquisision between groups receiving input through reading or RWL, and similar findings were reported by van Vu and Peters (2020).

Clearly, differing results have been found concerning the incidental acquisition of vocabulary as a result of reading, listening to, and reading-while-listening to texts. In order to shed more light on this issue, this study will investigate the incidental vocabulary learning outcomes of university students being exposed to texts in their L2 in the three

aforementioned manners. Thus, the main research question under consideration in this thesis is:

What are the differences in the incidental vocabulary learning outcomes of Dutch university students after reading, listening to, or simultaneously reading and listening to texts in English?

This will be done by providing the students with three different texts, one to be read, one to be listened to, and the last one to be read and listened to simultaneously. The three texts contain multiple exposures to target pseudowords, of which the participants will have to infer the meaning in order to comprehend the story. The number of words learned will be investigated, as well as differences in the type of word knowledge acquired. The types of word knowledge tested are form recognition, meaning recognition, and meaning recall. Additionally, the study will examine whether the different types of input lead to different degrees of retention of the vocabulary learned after a period of two weeks. In short, the sub-questions of this thesis’s main research question are:

1. What are the differences in the number of words acquired through the three different modes of input?

2. What are the differences in the types of word knowledge acquired through the three different modes of input?

3. What are the differences in the retention of the vocabulary items acquired through the three different modes of input?

In section 2 of this thesis the existing literature on incidental learning will be further explored. An overview of previous research into incidental vocabulary acquisition through reading, listening, and reading-while-listening (RWL) will also be given in this chapter. Section 3 will be used to give a description of the research method employed in this study, including information about the participants, the texts, target words, and vocabulary tests used. This section also includes a description of how the data was collected, as well as how the data was analyzed. In chapter 4, the results of the study will be presented. Section 5 contains a discussion and interpretation of the results, as well as the practical implications of the outcome of the study. This section also discusses the limitations of the current study. Finally, the conclusion can be found in chapter 6.

2. Background

2.1 Incidental vocabulary learning

The incidental learning of vocabulary occurs when learners acquire new L2 vocabulary items as a by-product of other activities in which the language is involved. Thus, it is the acquisition of L2 vocabulary without the conscious intention to learn new words (Laufer & Hulstijn, 2001). Incidental learning is often seen as the opposite of intentional learning, which could encompass the memorization of vocabulary lists or the completion of language exercises in the L2 (Barcroft, 2004). However, as suggested by Gass (1999), incidental and intentional learning do not necessarily have to be seen as opposites, but can also be taken as the two extreme ends of a continuum, along which the acquisition of some words can be more or less incidental than others (1999). Indeed, Huckin and Coady argue that “incidental learning is not entirely ‘incidental,’ as the learner must pay at least some attention to

individual words” (1999). Some degree of conscious effort is required to guess the meaning of the unknown words, which would conflict somewhat with the aforementioned definition of incidental learning. Nevertheless, although it should be recognized that acquiring vocabulary as a by-product of other activities can sometimes differ in the amount of conscious attention it involves, in this thesis it will be treated as a different process than intentional learning, as acquiring a word from context is very different from the intentional rote memorization of vocabulary lists. Thus, the definition of incidental learning as explained above will be maintained.

An early study showing that vocabulary can be acquired from context, without the learners having been instructed to pay attention to learning new words, was carried out by Nagy, Herman, and Anderson (1985). They found that school children who read texts for meaning showed gains in word knowledge of target vocabulary items from the texts (1985). Although this study only focused on learners reading and acquiring words in their native

language, similar results were found by Day, Omura, and Hiramatsu (1991) in an EFL context. In their study, Japanese learners of English who had read a short story were found to have incidentally acquired new English vocabulary from this activity (1991). While both of these studies target reading as an activity that can facilitate incidental vocabulary learning, which will be expanded on below in section 2.2, a large number of different activities and types of language input can lead to incidental acquisition. For example, Feng and Webb (2019) had groups of Chinese EFL learners consume the same information using different modes of input. The learners either read a transcript of a documentary, listened to the documentary, or watched it. It was found that all three modes of input led to English

vocabulary gains, and that these gains were retained one week later (2019). Feng and Webb (2019) did not find any differences in the effectiveness of the different input types, something which will be explored further in section 2.4 and 2.6 of this paper. Yang and Sun (2013) found gains in English vocabulary knowledge for Taiwanese EFL learners who watched online lectures in their L2. Lastly, language learning in immersion settings depends on the incidental acquisition of the L2 through listening to classroom instruction and through interaction with teachers and classmates. This approach has been found to enable learners to attain high levels of proficiency in the target language (Genesee, 1987). Thus, written, audio, and audiovisual input, and interaction with other speakers can all effectively lead to

incidental vocabulary learning. In this paper, only three modes of input will be focused on, namely: reading, listening, and reading-while-listening (RWL). More studies investigating these specific modes will be examined below, but the literature discussed so far shows ample evidence that incidental vocabulary learning can and does occur.

Not only has incidental acquisition been found to be a real and effective method of vocabulary learning, some go as far as to say incidental learning is superior to intentional learning. Ahmad (2012) compared ESL learners in Saudi Arabia who had learned new words

through incidental and intentional learning, and found an advantage for incidental vocabulary learning. He argues that this advantage stems from the fact that incidental acquisition requires more cognitive resources, as learners need to take all context clues into account in order to guess the meaning of the unfamiliar word, whereas intentional learning tends to consist of mere rote learning, without any deeper mental processing (Ahmad, 2012). This view is supported by results from a study by Hulstijn (1992), who found that L2 learners who had guessed the meaning of unknown words from context retained these words better than learners who had been provided with synonyms of the same words. Despite these results pointing to an advantage for incidental learning, there is also evidence to the contrary, as Paribakht and Wesche (1997) found that words learned intentionally by using them in exercises were retained better than those learned incidentally by inferring the meaning of the word from context.

It has by now been established that incidental learning is an effective way to acquire vocabulary, but there remains the question of how incidental vocabulary learning occurs. After all, not every new word an L2 learner encounters actually gets learned, for instance, ESL students in a study by Paribakht and Wesche (1999) ignored approximately half of the unfamiliar words they encountered in a text. For the other half, their main strategy was to inference the meaning of the words. Incidental vocabulary learning largely relies on inferencing: learners need to use cues from the context, and make guesses about what the unknown word could mean on the basis of that information (Paribakht & Wesche, 1999). However, once the learner has correctly inferenced the meaning of the word, the word is not automatically learned. The acquisition of a new word is gradual and incremental, and

repeated exposure is required (Gass, 1999). Paribakht and Wesche (1999) also stress that multiple encounters with a word are necessary to develop a mental representation of the word, and store its form and meaning in long-term memory. No real agreement has been

reached of exactly how many exposures to a new word are needed to acquire it, and estimates vary from only 1 to as many as 16 exposures (Huckin & Coady , 1999). Moreover, this amount could vary both between different learners (differences in aptitude, age, etc.), and between different words (some word meanings could be more difficult to inference, or conversely, the word form may be very close to an L1 cognate, making it easier to learn).

Models of second language acquisition, even when they are not specific to incidental learning per se, can provide a lens through which to view incidental vocabulary acquisition. Gass (1988) has proposed a framework of L2 learning that describes how learners go from receiving L2 input to using the linguistic forms in that input (in this case vocabulary items) in their own L2 output. The first step in this process is apperceived input, in this stage the learners notices the new linguistic form (Gass, 1988). In incidental vocabulary learning, this would be when learners notice an unfamiliar word in the input. The second step is that of

comprehended input, which is when the learner understands the meaning of the input,

including its unfamiliar items (Gass, 1988). For incidental vocabulary acquisition, this is where inferencing comes into play: the learner needs to guess the meaning of the novel word they have noticed in order to comprehend the input. The comprehended input can then potentially lead to intake, though not all input becomes intake (Gass, 1988). Intake refers to the process of input becoming part of the learner’s linguistic knowledge (Gass, 1988). Thus, for vocabulary acquisition, the learner is forming a long-term memory representation of the word’s form and meaning. When, over time, the learner has received enough input and has gathered enough information about the new item, intake leads to the integration of the new linguistic form into the L2 language system (Gass, 1988). In incidental vocabulary

acquisition, this is when the lexical entry of the previously unfamiliar vocabulary item has been established. The final stage is output, when the learner is able to use the new item in their own L2 production (Gass, 1988).

A somewhat similar model was proposed by Ma (2014). However, whereas Gass’s (1988) framework is general to all second language learning, Ma specifically focuses on vocabulary acquisition. Ma’s model consists of four stages: (1) perceiving the word form, (2) assessing the word meaning, (3) building the word entry, and (4) retrieving the word (2014). Stage 1 is very similar to Gass’s (1988) concept of apperceived input; the learner notices an unknown vocabulary item in the input (Ma, 2014). During stage 2 the learner obtains the meaning of the novel word, for example by looking it up in a dictionary, or, as is most likely in the case of incidental acquisition, through inferencing (Ma, 2014). Stage 3 refers to the repeated exposures needed to strengthen the new word’s lexical entry, each encounter of the word helps the learner connect its form and meaning (Ma, 2014). Lastly, in stage 4 the learner is already using the new vocabulary item in their L2 output, and each time the item is retrieved from the learner’s lexicon the memory trace gets strengthened and the lexical entry is further consolidated (Ma, 2014). Both of the models discussed in this section can be useful in the interpretation of empirical data on incidental vocabulary learning.

2.2 Reading

One of the modes of input that can facilitate incidental vocabulary acquisition is reading. The majority of incidental learning focuses on this type of input, as can be seen in the studies discussed above, such as the study by Day, Omura, and Hiramatsu (1991) in which they found evidence for incidental vocabulary acquisition through reading by Japanese EFL learners. In a literature review of existing studies on incidental acquisition, Restrepo Ramos (2015) concludes that “there is strong evidence that supports the occurrence of

incidental vocabulary learning through reading for meaning comprehension.” Indeed, there is a plethora of studies that support this, another example being one by Ponniah (2011). In this study, two groups of adult ESL students in India were compared: one group read a short story containing unfamiliar words for meaning, and the other groups studied the dictionary

definitions of these words, without reading the story (Ponniah, 2011). A vocabulary test showed that the students who intentionally studied the new vocabulary items could give the word definitions, but were not able to use the words in a sentence (Ponniah, 2011). On the other hand, the group who had read the story was able to do both, through incidental vocabulary acquisition (Ponniah, 2011). Interestingly, before participating in the study, the students had reported that they did not believe they could learn vocabulary by simply reading, but they still successfully acquired the new words in this manner (Ponniah, 2011). Tekmen and Daloglu (2006) investigated incidental vocabulary acquisition through reading in Turkish learners of English at three proficiency levels: intermediate, upper intermediate, and

advanced. After reading a text, all groups showed significant vocabulary gains, but the students with higher proficiency levels acquired more words than those with a lower level (Tekmen & Daloglu, 2006). The reason for this could be that these more advanced learners are better able to comprehend the text, as a reader needs to already know between 95% and 98% percent of the words in a text to fully understand it (Restrepo Ramos, 2015). Having a better understanding of the surrounding context may have led the high proficiency learners to be better at inferencing the meaning of the target words, and thus acquiring them. When the lower proficiency level learners encountered unknown words that they could not guess the meaning of, because they could not comprehend the context, they probably simply ignored the words, which is what participants in the previously discussed study by Paribakht and Wesche (1999) also did. In a meta-analysis of 20 studies on incidental vocabulary

acquisition, it was found that learners only learn approximately 15% of the new vocabulary items they encounter during normal reading (Swanborn & de Glopper, 1999). In order to shed more light on what exactly happens when L2 learners read a text, recent incidental

acquisition studies have started using eye-tracking technology. Mohamed (2018) had L2 English learners with a range of L1 backgrounds read a text with 20 target words replaced by

pseudowords. The eye-movement data showed learners spent more time looking at the unknown pseudowords than on familiar words, and that this amount of time decreased with repeated exposure to the word (Mohamed, 2018). The students did learn the target words to some extent, as the average vocabulary gains were 41.8% for form recognition, 30.3% for meaning recognition, and 12.9% for meaning recall (Mohamed, 2018). The amount of time spent on a particular word predicted the student’s success in acquiring this word, showing that the more attention a learner pays to a particular word, the more likely they are to learn it (Mohamed, 2018).

2.3 Listening

Listening is another mode of input that makes it possible for learners to incidentally acquire vocabulary. For example, Pavia, Webb, and Faez (2019) found that just listening to music could lead to acquisition, as participants in their study showed vocabulary gains after listening to songs, especially after listening to the same song multiple times. However, learners can listen not only to music, but also to stories or to readings of written text, which contain more varied and difficult vocabulary items than most songs, and would thus be more useful for L2 vocabulary learning. Lenhart, Lenhard, Vaahtoranta, and Suggate (2018)

investigated the incidental vocabulary acquisition of German kindergarten-aged children after listening to stories. Learning gains were found both when stories were read aloud or told freely (Lenhart, Lenhard, Vaahtoranta, & Suggate, 2018). Elley (1989) found that elementary school children in New Zealand showed vocabulary gains of 15% from listening to a story, and that the gains increased to 40% when the children received teacher explanation about the target words. This advantage is likely due to the fact that the teacher explanations caused the children to pay more attention to the target words. While both of these studies show the potential of listening for incidental acquisition, they focus on vocabulary acquisition in the L1. Al-Shehri (2015) researched the incidental vocabulary acquisition of children in an L2. In

this study, children who were L1 Arabic speakers and EFL learners listened to stories with target pseudowords, and were found to be able to acquire these words (Al-Shehri, 2015). It was also found that multiple exposures and explanations about the new words increased the amount of acquisition (Al-Shehri, 2015), in accordance with Elley’s findings (1989). A study that investigates incidental L2 vocabulary acquisition through listening by learners much older than the ones in the studies discussed so far was carried out by Ellis and Le (2016). They let Chinese university students listen to texts in English, and found that this facilitated the acquisition of L2 vocabulary (Ellis & Le, 2016). The results also showed that listening to the text multiple times led to higher vocabulary gains, providing further support for the claim that multiple exposures are beneficial for incidental vocabulary learning (Ellis & Le, 2016). Since inferencing the meaning of a new word is an important part of incidental acquisition, Ellis and Le (2016) provided some of the learners with inference-training. However, these students did not perform better than those who had not received the training (Ellis & Le, 2016). Thus, learners clearly do not need any kind of training in order to be able to learn vocabulary incidentally. Van Zeeland and Schmitt (2013) also investigated the incidental vocabulary acquisition by university students, and included English learners with different L1 backgrounds in their study. They aimed to differentiate the types of word knowledge L2 learners can acquire through listening (van Zeeland & Schmitt, 2013). The learners listened to four English texts which contained 24 target pseudowords (van Zeeland & Schmitt, 2013). They were then tested on three types of vocabulary knowledge: form recognition, grammar recognition (knowledge of the word’s part of speech), and meaning recall (van Zeeland & Schmitt, 2013). The scores showed that students performed best on the form recognition test, slightly lower on grammar recognition, and much lower on meaning recall (van Zeeland & Schmitt, 2013). This implies that when a learner encounters a new word, the word form gets learned earlier than the word meaning, and that more exposures are needed to connect a

word’s form and meaning than to recognize its form. Thus, these findings provide more insight into exactly how learners acquire new words by listening to L2 input.

2.4 Comparing reading and listening

The research findings discussed in the preceding sections show clear evidence that both reading and listening can lead to incidental vocabulary acquisition. However, it is possible that one type of input is more beneficial than the other, which is a question that has been investigated in a number of studies comparing the two input modes. Suggate, Lenhard, Neudecker, and Schneider (2013) found that second and fourth grade children in Germany learned more words from listening to someone telling a story, than from independently reading this story themselves. Similarly, Geva, Galili, Katzir, and Shany (2017) found that Hebrew-speaking fourth grade children were better at inferencing the meaning of an

unknown word when they encountered this word during listening than during reading. These two studies would suggest an advantage for listening over reading. However, both studies deal with very young participants, and more importantly, focus on L1 rather than L2 learning. Other studies have compared these input modes for L2 learners. For example, in the

previously discussed study by Feng and Webb (2019) Chinese university students learning English were exposed to the same text by either reading, listening, or viewing. Vocabulary gains were found for all groups, but no differences in vocabulary between the different input modes were found, suggesting that the type of input has no influence on incidental

vocabulary acquisition (Feng & Webb, 2019). However, Hatami (2017) conducted a study with similar participants (Farsi university students learning L2 English) and found that students who had read a text containing 16 pseudowords scored higher than students who had listened to the same text on a vocabulary post-test. The vocabulary test included five different types of vocabulary knowledge: spoken form, written form, part of speech, syntagmatic association, and form-meaning connection (Hatami, 2017). The reading group outperformed

the listening group on all five vocabulary knowledge measures on the immediate post-test, and also attained higher scores on a delayed post-test three weeks later (Hatami, 2017). Thus, this study points to strong advantages for reading over listening as a means of incidental learning. Chang and Ma (2018) also investigated the incidental acquisition of L2 English vocabulary by Chinese students. They investigated three types of vocabulary knowledge: form, meaning, and production (Chang & Ma, 2018). Their results showed that overall, reading resulted in more vocabulary acquisition than listening, but reading and listening led to similar results with regard to knowledge about the meaning of a vocabulary item (Chang & Ma, 2018). Thus, while reading was again found to be more beneficial for incidental

acquisition than listening to a text, this advantage may not be present in every type and every stage of word learning. Lastly, Vidal (2011) compared the incidental L2 English vocabulary acquisition by Spanish students of different proficiency levels after reading or listening to three texts. Again, greater vocabulary gains were found for the groups who had read rather than listened, and this was the case for all proficiency levels (Vidal, 2011). The students who had read the texts were also better able to retain the acquired vocabulary (Vidal, 2011). However, as proficiency increased, the difference in vocabulary gains between reading and listening groups decreased (Vidal, 2011). This suggests that for higher proficiency learners, the mode of input may not matter as much as it does for beginning L2 learners. Overall, while some L1 studies seem to point to advantages for listening (Suggate et al., 2013; Geva et al., 2017), and some L2 research has not found differences between the two modes of input (Feng & Webb, 2019), the vast majority of research findings point to reading being more facilitative to incidental vocabulary acquisition than listening (Hatami, 2017), although this difference may not be as pronounced for all types of vocabulary knowledge (Chang & Ma, 2018) and for all levels of learners (Vidal, 2011).

2.5 Reading-while-listening

The studies discussed so far have demonstrated the value of both reading and listening as modes of input for incidental vocabulary acquisition. However, another option would be for learners to read texts while also listening to that same text being read out simultaneously. In this paper, this mode of input will be referred to as reading-while-listening (RWL). Input of this type could be even more beneficial for word learning than merely reading or listening, since both the orthographic and phonological form of the word are provided in the input (Valentini, Ricketts, Pye, & Houston-Price, 2018). RWL is a practice currently often used for learners with reading problems, and has been found to lead to increases in reading fluency for adults with reading skill deficits (Winn, Skinner, Oliver, Hale , & Ziegler , 2006), and to be beneficial to the reading comprehension of English texts by Slovenian EFL learners with dyslexia (Košak-Babuder, Kormos, Ratajczak, & Pižorn, 2019). It has also been found to be beneficial for L2 learners in other aspects of learning, as Chang and Millett (2014)

investigated how extensive RWL affected listening fluency for EFL university students, and found that students who participated in RWL made larger improvements in listening fluency than those who had done extensive reading or extensive listening, using the same texts. Milliner (2019) also found gains in performance on both listening and reading tests by Japanese university learners of L2 English after participating in extensive reading-while-listening. After RWL, these students also scored better on a vocabulary test, suggesting that incidental vocabulary acquisition was facilitated by the RWL activities (Milliner, 2019). Chang (2011) also investigated vocabulary gains by students who had taken part in extensive RWL. In this study, Taiwanese secondary school students enrolled in an English listening course were divided into two groups: one group spent their learning time reading books while simultaneously listening to audiobooks, while the other group received formal listening instruction (Chang, 2011). After 26 weeks, the RWL students outperformed the control group

on listening fluency, but more importantly, had made much larger vocabulary gains than the group receiving formal instruction (Chang, 2011). This demonstrates that RWL is highly effective for incidental vocabulary acquisition. A similar study was carried out by Webb and Chang (2012), as they also investigated incidental vocabulary learning through RWL by Taiwanese secondary school students learning English. However, in this study, they did not compare the RWL group to a group receiving formal instruction, but to a group that did extensive reading (Webb & Chang, 2012). Both groups in the study read 28 short texts, and both groups received multiple exposures to each text, but whereas one group only read the texts, the other group read them while also listening to a recording of the text (Webb & Chang, 2012). Webb and Chang (2012) found that learners in both conditions made vocabulary gains, but that RWL led to higher gains than reading only. This suggests that RWL may be superior to reading as a mode of input for incidental vocabulary acquisition. Further studies comparing these two input modes, along with listening as a third input option, will be discussed below.

2.6 Comparing RWL to reading and listening

A number of different studies have compared reading-while-listening to other input modes for incidental vocabulary acquisition. Valentini, Ricketts, Pye, and Houston-Price (2018), although investigating vocabulary acquisition by young L1 English learners, found some differences in vocabulary learning between reading, listening, and RWL. Their results showed that phonological learning can occur with any type of input, even when there is no phonological representation of the word in the input, as is the case with reading Valentini et al., 2018). Orthographic learning could only occur through reading or RWL, and semantic learning occurred in all conditions, but was greatest when the mode of input was RWL (Valentini et al., 2018). Tragant Mestres, Llanes Baró, and Pinyana Garriga (2019) compared only reading and RWL as input modes, and also included a control group in their study of

incidental L2 English vocabulary learning by young Spanish speakers. It was found that both the reading and RWL group had higher vocabulary gains than the control group, showing the effectiveness of both input types (Tragant Mestres et al., 2019). However, no significant difference was found between the two experimental groups, suggesting no advantage for one type of input over the other (Tragant Mestres et al., 2019). Reading and RWL were also compared by van Vu and Peters (2020) with reading with textual input enhancement (bolding or underlining target words) as an additional type of reading input under investigation. For a group of Vietnamese adult EFL learners, all three types of input led to incidental vocabulary acquisition, and the highest gains were attained with reading with textual input enhancement (van Vu & Peters, 2020). However, in accordance with the findings by Tragant Mestres et al. (2019), no differences in vocabulary gain between reading and reading-while-listening were found. These two studies suggest that there are no differences in incidental acquisition outcomes between reading and RWL. However, in the aforementioned study by Webb and Chang (2012) learners in an RWL condition made larger vocabulary gains than learners who only read texts. This is supported by Teng (2018). Teng (2018) let Chinese university

students learning English as a foreign language either read a text, or read it while simultaneously listening to the text. Significant vocabulary gains were found in both conditions, as long as exposure to the target items was frequent enough, but learners in the RWL conditions learned more words than those in the reading-only condition (Teng, 2018). Thus, while some research has found no differences between reading and RWL for incidental acquisition, other findings show advantages for RWL over simply reading. The study that is perhaps most similar to the present study is Brown, Waring, and Donkaewbua (2008), as it includes the same three types of input for incidental vocabulary acquisition: reading,

listening, and reading-while listening. In this study, Japanese students were exposed to three stories containing target pseudowords, by either reading them, listening to them, or reading

while listening to them (Brown et al., 2008). A meaning-translation test administered

immediately after the exposure to the input showed that students had acquired the most words in the RWL condition, followed by the reading condition, and that the smallest amount of words was learned in the reading condition (Brown et al., 2008). Post-tests also revealed that after three months, the students could recall hardly any of the acquired words, but that words with a higher frequency in the text were the most likely to be remembered over a longer period of time (Brown et al., 2008). This outcome again shows the importance of repeated exposure to new vocabulary items for incidental acquisition. The finding that listening leads to the least vocabulary learning is in accordance with most findings discussed in section 2.4, which found that reading leads to more incidental vocabulary acquisition than listening (Hatami, 2017; Chang & Ma, 2018; Vidal, 2011). On the other hand, the finding than RWL is most beneficial for incidental acquisition corresponds to some of the research comparing RWL to other input modes (Webb & Chang, 2012; Teng, 2018), while others claim that there is no difference in incidental vocabulary acquisition outcomes between reading and RWL (Tragant Mestres et al., 2019; van Vu & Peters, 2020). Clearly, results regarding this question have been mixed, and further research is needed.

2.7 Statement of purpose

Vocabulary learning is an extremely important part of second language acquisition, and as the studies discussed above show, incidental learning can play a large part in an L2 learner’s vocabulary acquisition. For L2 learners as well as instructors, it would be useful to know how to optimize L2 exposure in order to best facilitate incidental vocabulary

acquisition. For this, it would be beneficial to know how different modes of input affect incidental learning. Although previous studies have investigated this, the results have been mixed. Most results point to an advantage for reading over listening (Hatami, 2017; Chang & Ma, 2018; Vidal, 2011), but while some report a similar advantage of reading-while-listening

over reading (Webb & Chang, 2012; Teng, 2018), other findings seem to suggest no difference between the input modes (Tragant Mestres et al., 2019; van Vu & Peters, 2020). With the aim of gaining a better understanding of the differences between input modes for incidental vocabulary acquisition, the current study will investigate the following research question:

What are the differences in the incidental vocabulary learning outcomes of Dutch university students after reading, listening to, or simultaneously reading and listening to texts in English?

This will be investigated by exposing Dutch learners of L2 English, all university students at the time of the experiment, to three English texts. The participants will read one text, listen to another one, and the last text will be read while simultaneously listening to a recording of the story. Each story contains five different target pseudowords, which all appear exactly five times in the text. The learners then take a vocabulary test, where three types of word knowledge are measured: form recognition, meaning recognition, and meaning recall. Two weeks after the input exposure and the immediate post-test, a delayed post-test measuring the same knowledge will be administered.

In order to arrive at a detailed answer to the main research question, three sub-questions will be taken into consideration. The first sub-question is:

1. What are the differences in the number of words acquired through the three different modes of input?

Based on the existing literature, it is expected that the listening condition will lead to the least amount of incidental vocabulary learning. An advantage for RWL over reading and listening is also expected.

2. What are the differences in the types of word knowledge acquired through the three different modes of input?

The three different types of word knowledge that will be compared are form recognition, meaning recognition, and meaning recall. It is predicted that all types of input can lead to form recognition of the unknown words. However, since meaning recall requires the most robust form-meaning connection for the new word, and since RWL facilitates the creation of a strong lexical entry, because the learner is exposed to both the phonological and

orthographic form of the novel vocabulary item, it is expected that RWL will lead to more meaning recall than reading or listening.

The last sub-question is:

3. What are the differences in the retention of the vocabulary items acquired through the three different modes of input?

Previous studies have found higher retention of the incidentally acquired vocabulary for reading over listening (Hatami, 2017; Vidal, 2011). Since reading has overall been found to be more conducive to incidental learning than listening, it follows that this same advantage likely extends to retention. Since RWL is hypothesized to lead to more incidental vocabulary acquisition than reading and listening, RWL is also expected to result in the highest retention of the newly acquired lexical items.

3. Method

In order to answer the research question, this study tested the incidental L2 vocabulary acquisition of adult Dutch EFL learners through different modes of input. The learners were exposed to English short stories in three different input conditions, reading, listening, and RWL. The three stories all contained 5 target words, which appeared in the text exactly five times. The target words were not existing English words, but pseudowords, to ensure that the

learners were not previously exposed to and had no pre-existing lexical knowledge of the target words. A post-test was administered directly after the input, measuring three types of vocabulary knowledge: form recognition, meaning recognition, and meaning recall. Two weeks later, a delayed post-test was administered to measure the retention of these same three types of vocabulary knowledge. Thus, this study examines the influence of mode of input on incidental vocabulary learning outcomes, by testing the acquisition and retention of different types of vocabulary knowledge of target pseudowords.

3.1 Participants

This study included thirty-one participants, who were all students at a Dutch university enrolled in the same second year course in linguistics, and participation in the experiment was a compulsory assignment in the course. Students in the course who did not speak Dutch were excluded from the experiment. Thus, all participants had a Dutch

nationality, and were native speakers of Dutch. Twenty-five of the participants considered Dutch to be their L1, but six participants had grown up speaking other languages alongside Dutch (five spoke Frisian, and one spoke Farsi). The participants spoke English as an L2, and since they were all students of English at university, they had a high L2 proficiency level. Twenty-one female, and ten male participants were included in this study, and all participants were between the ages of 19 and 25, with the mean age being 21.06. The students were divided into three groups in order to eliminate text effects, by varying which mode of input was used for which story between the groups. Students were randomly assigned to one of the three groups. All groups were meant to be of the same size, but since some students failed to participate this was not the case. Group 1 consisted of eleven participants, and so did group 2, but group 3 had only 9 participants. Nevertheless, even though the groups did not have the exact same amounts of students, the groups were still of similar sizes. Two of the thirty-one

participants did not participate in the delayed post-test, leaving twenty-nine participants for this test.

3.2 Materials

Before being exposed to the input, participants were asked to complete a short survey. This survey included questions about the participant’s age, gender, nationality, first language, and course of study. The survey also asked for the participant’s student number, which was used to match the data entries of the immediate and delayed post-tests, and to check which students had completed their participation in the experiment, as this was a compulsory part of the course they were enrolled in. However, after this, the student number was removed from the data, and the data was processed anonymously. The survey questions can be found in Appendix A.

The texts that were used as input in this experiment were excerpts from short stories, taken from a short story collection for young adults (Chapman, 2019). This collection was chosen because the younger target audience meant that the language and the content would not be extremely complicated, so that it could be ascertained that all participants

comprehended the content of the stories. The short stories also all revolved around food and restaurants, which is a topic all participants were likely familiar with and interested in. Three stories were selected from the collection, and excerpts were taken from the beginning of the stories. The excerpts were then adapted to remove potentially difficult vocabulary items, and to shorten them to approximately the same length. Text 1 consisted of 1035 words, text 2 of 1239, and text 3 of 1226. From each text, five words were then chosen to be replaced by pseudowords. Different word types were chosen, including nouns, verbs, adjectives, and a preposition. These words were then replaced by English-sounding pseudowords generated using the ARC Nonword Database (Rastle, Harrington, & Coltheart, 2002). All pseudowords replacements were monosyllabic, to make the form of each word equally difficult to learn.

The 15 target words and the existing words that they replaced can be found in Table 1. In order to ensure that all target words occurred exactly five times in their respective texts, some instances of the target word were rephrased or deleted from the text, and sentences were added to the stories to facilitate an additional occurrence of other target words. In many cases, the original sentences of the stories were expanded to make the meaning of a target word more evident from the context. The adapted excerpts were read by two Dutch speakers with a high L2 English proficiency, and any lexical items that were unknown to them were taken out of the text, and unclear parts were rewritten. The pilot readers were also asked to inference the meaning of the target words, in order to confirm that the meaning of the pseudowords could be learned from the context. The three adapted texts can be found in Appendix B.

Table 1. Original target words and pseudoword replacements per text.

Text 1 Text 2 Text 3

Original Pseudoword Original Pseudoword Original Pseudoword

Street Chasp New Proothe Restaurant Smusk

Father Zeak Kitchen Wheech Phone Jalp

Food Bloin Hear Flooge Before Sprowse

Rain Mipe Eat Hisp Face Chisk

Sad Dwilk Prize Shomp Ask Ress

For the listening condition, a recording of each of the story excerpts was made. The texts were read out by a female native speaker of English, speaking in a General American

accent. The recordings were made using a smartphone. The recording of text 1 was 8 minutes and 13 seconds long, text 2 was 9 minutes 19 seconds, and text 3 8 minutes 38 seconds.

To test the acquired vocabulary knowledge after exposure to the input, an immediate post-test was used. There was a separate test for each of the three texts. The tests consisted of four different parts: text comprehension, form recognition, meaning recognition, and meaning recall. Text comprehension was included to ascertain that each participant had read and understood the text before taking the post-test, and that participants who did not really read the text could potentially be excluded from the results. This part consisted of three open questions about events from each story, that the participants were asked to answer in one phrase or sentence. The text comprehension questions were answered sufficiently by all participants.

The form recognition test consisted of the five target words, along with five distractor pseudowords, also generated with the ARC Nonword Database (Rastle, Harrington, &

Coltheart, 2002). For each word, the participant was asked to answer whether or not the word occurred in the text they had just read, the answer options being “yes” and “no”. The ten words were presented in a randomized order.

The meaning recognition test included a multiple-choice question for each of the five target words from a text. Participants were asked to choose the word whose meaning was the closest to the target word out of four options. The three distractor options were always the same part of speech as the correct answer, and they were semantically related to the actual meaning. Both the order of the questions and the order of the answer options was

randomized.

For the meaning recall part of the post-test, participants were simply asked to give the meaning of the target word by typing out the Dutch translation of the word. A large number of students actually answered this question by giving the meaning in English rather than

Dutch. These answers were also accepted as correct, as they displayed that the participant had learned the meaning of the target word. Again, the order of the words was randomized.

The delayed post-test did not include the text comprehension questions, but did include all other parts of the immediate post-test for each of the three texts. The meaning recognition and meaning recall tests were kept exactly the same. However, to prevent

students from thinking they recognized distractor from the text, when in fact they recognized them from the immediate post-test’s form recognition questions, new distractor words were generated for use in the form recognition parts of the delayed test. In the delayed post-test, all vocabulary knowledge tests for the target words from one text were administered for before moving on to the same vocabulary tests for another of the texts. The order of the three parts of the post-test (for the three different texts) was randomized between participants. All parts of both the immediate and delayed post-tests can be found in Appendix C.

3.3 Procedures

All parts of the experiment were carried out online, on the participant’s own devices, at a time of their choosing. This method was chosen because at the time of the experiment, all schools and universities in the Netherlands were closed as part of pandemic response

measures, making it impossible to carry out the experiment in the classroom. The participants received an e-mail with links to the experiment, and could carry it out at a time of their own choosing, within a time frame of five days. The e-mail contained links to three separate surveys constructed in Qualtrics (Qualtrics, Provo, UT), for each of the three conditions, and the participants could complete them in any order and with as much time in between as they wished. Each of the three parts of the experiment started with the short survey on the

participant’s basic background information, as described above. Before being exposed to the input, the students were told they were about to read, listen to, or read and listen to a short English story, containing unfamiliar words. The number of unfamiliar words was not

specified. The students were instructed to read for meaning, and were informed that they were not allowed to take notes. In the reading condition, the text was then displayed, and the students were asked to read the whole story before moving to the next part of the survey. In the listening condition, a media player with the audio file was displayed on the page, and the participants were asked to listen to the entire recording. For the RWL condition, both the media player and the text were displayed, and students were explicitly instructed to utilize both types of input, and make sure they were reading the text in time with the recording, without reading ahead. For the three different groups, different texts were used for each of the input conditions. The distribution of the texts across the input conditions for each group can be found in Table 2.

Table 2. Distribution of the texts over input conditions.

Reading Listening RWL

Group 1 Text 1 Text 2 Text 3

Group 2 Text 3 Text 1 Text 2

Group 3 Text 2 Text 3 Text 1

After the input phase, the students moved directly to the immediate post-test for that text. Thus, they completed the post-test of one text before going through the input stage of the next text. The text comprehension questions were presented first, followed by form recognition, meaning recall, and meaning recognition. Form recognition was tested first, because if the participants had already seen which words were tested in the other parts, they would know which of the words were target words. Meaning recall was positioned before meaning recognition so that participants would not recall the meaning of the words from the multiple-choice options, rather than inferencing the meaning from the stories. The link to the

delayed post-test was sent out exactly 2 weeks after the first e-mail, and the participants were again given five days to complete it. All questions and instructions in the survey, immediate post-test, and delayed post-test were in Dutch.

3.4 Analysis

The three different parts of the vocabulary post-test were scored by adding up the number of correctly answered items. Besides the three separate scores, an overall score was calculated by taking the average of the scores on the three parts. The average rather than the sum of the three scores was used because the form recognition test consisted of ten items, while meaning recognition and meaning recall had only five items, which would have given form recognition more weight than the other parts if the sum of the scores was used. To compare the three different input modes, a within-subject design was used. Each participant had been exposed to texts in all three conditions, and the vocabulary scores of all participants for the reading condition, listening condition, and RWL condition could be compared. The independent variable was thus the input condition, with three levels: reading, listening, and RWL. The dependent variables were the eight different vocabulary scores (overall, form recognition, meaning recognition, and meaning recall, on the immediate and the delayed post-test) for each condition. The differences between the vocabulary scores on the

immediate and delayed post-tests were determined by subtracting each participant’s delayed post-test score from the corresponding immediate post-test score. Since the data was not normally distributed, the non-parametric Friedman test was used to detect differences in vocabulary scores between the conditions. Pairwise comparisons were carried out using a paired Wilcoxon signed-rank test, with the Bonferroni multiple testing correction method. Effect sizes were determined using Kendall’s W. The alpha level for all statistical tests was set at p < 0.05, and all tests were carried out using R (R Core Team, 2018) and RStudio (RStudio Team, 2016).

4. Results

4.1 Overall vocabulary scores

To investigate the effectiveness of reading, listening, and RWL as modes of input for incidental L2 vocabulary acquisition, the overall vocabulary scores on the immediate post-test were compared. The mean, minimum, and maximum scores for each condition can be found in Table 3, along with the standard deviations.

Table 3. Descriptive statistics for the overall vocabulary scores obtained on the immediate

post-test in the reading, listening, RWL conditions.

Input condition Mean Min Max SD

Reading 5.87 3.67 6.67 0.83

Listening 4.96 2 6.67 1.13

RWL 5.96 4 6.67 0.69

A Friedman test found significant differences between the overall vocabulary scores for target words acquired in the three different input conditions, χ2(2, 30) = 18.22, p < 0.001. Kendall’s W revealed this to be a small effect, W = 0.29. A paired Wilcoxon signed-rank test found no significant differences between the overall scores obtained in the reading and RWL conditions, but a significant difference was found between reading and listening conditions (p < 0.001), and between RWL and listening (p < 0.001). The overall scores for all three

Figure 1. Boxplot showing the overall vocabulary scores on the immediate post-test in the

reading, listening, and RWL conditions.

As can be seen in Figure 1, the lowest overall scores were obtained in the listening condition. Relatively high scores were obtained in both the reading and RWL conditions, with slightly higher scores for the RWL condition, though this difference was not found to be significant. Thus, the overall vocabulary scores show that listening as a mode of input leads to significantly less incidental vocabulary acquisition than reading or RWL.

4.2 Scores per vocabulary knowledge type

To see how input condition affects the acquisition of different types of vocabulary knowledge, the scores on the three individual parts of the vocabulary test were compared between conditions. The three types of vocabulary knowledge that were tested are form recognition, meaning recognition, and meaning recall.

Table 4. Descriptive statistics for the form recognition scores obtained on the immediate

post-test in the reading, listening, RWL conditions.

Input condition Mean Min Max SD

Reading 9.52 7 10 0.77

Listening 8.87 4 10 1.52

RWL 9.68 9 10 0.48

Using a Friedman test, a significant difference between the form recognition scores in the three conditions was found, χ2(2, 30) = 6.32, p = 0.042. Kendall’s W showed that this difference is of a small effect size, W = 0.10. A paired Wilcoxon signed-rank test found no significant differences between the form recognition scores in the reading and RWL conditions, and the reading and listening conditions. However, a significant difference was found between the form recognition scores in the listening and RWL conditions, p = 0.025. The form recognition scores are also visualized in Figure 2.

Figure 2. Boxplot showing the form recognition scores on the immediate post-test in the

reading, listening, and RWL conditions.

As can be seen in both Table 4 and Figure 2, the form recognition scores were very high in all input conditions. The highest scores were obtained in the RWL condition, although the mean score is very close to that of the reading condition. While the form recognition scores were still high, they were lower for the listening condition than the other conditions, which is reflected in the statistically significant difference that was found between the form recognition scores obtained in the listening and RWL conditions.

Table 5 shows the mean, minimum, maximum, and standard deviation of the meaning recognition scores.

Table 5. Descriptive statistics for the meaning recognition scores obtained on the

immediate post-test in the reading, listening, RWL conditions.

Input condition Mean Min Max SD

Reading 4.65 2 5 0.75

Listening 4 1 5 1.10

RWL 4.68 2 5 0.70

A significant difference between the meaning recognition scores of the three input conditions was found using a Friedman test, χ2(2, 30) = 14.18, p < 0.001. Kendall’s W found this to be a small effect, W = 0.23. A paired Wilcoxon signed-rank test found a significant difference between the meaning recognition scores attained in the listening and RWL conditions, p = 0.01, and between the reading and listening conditions, p = 0.006. The difference between the scores in the reading and RWL conditions was not found to be significant. The meaning recognition scores are visualized in Figure 3.

Figure 3. Boxplot showing the meaning recognition scores on the immediate post-test in the

reading, listening, and RWL conditions.

As Figure 3 shows, the meaning recognition scores were very high for the target words the participants were exposed to in the reading and RWL conditions, with a large number of participants scoring the maximum amount of points. In the listening condition, the scores are more spread, with some participants scoring as high as those in the other

conditions, and other participants scoring much lower. This resulted in the significant

differences between the meaning recognition scores in the listening condition and those in the other conditions.

Descriptive statistics for the meaning recall scores in the three conditions are displayed in Table 6.

Table 6. Descriptive statistics for the meaning recall scores obtained on the immediate

post-test in the reading, listening, RWL conditions.

Input condition Mean Min Max SD

Reading 3.45 0 5 1.36

Listening 2 0 5 1.61

RWL 3.52 0 5 1.34

A Friedman test found significant differences between the meaning recall scores attained in the three different input conditions, χ2(2, 30) = 19.02, p < 0.001. Using Kendall’s

W, this was found to be a moderate effect, W = 0.31. A paired Wilcoxon signed-rank test did

not find the difference between the scores in the reading and RWL conditions to be statistically significant. The differences between reading and listening, p < 0.001, and

listening and RWL, p < 0.001, did appear to be significant. Figure 4 shows a visualization of the meaning recall scores obtained in the reading, listening, and RWL conditions.

Figure 4. Boxplot showing the meaning recall scores on the immediate post-test in the

reading, listening, and RWL conditions.

There were clear differences in the meaning recall scores between the three input conditions, as is evident from Table 6 and Figure 4. Similar to the other types of vocabulary knowledge, the scores were highest in the RWL condition, closely followed by reading, and were the lowest in the listening condition. This is also reflected in the statistical significance of the differences between the listening condition meaning recall scores and the scores in the other conditions.

There are similarities and differences between the effects of input condition on the acquisition of the different types of vocabulary knowledge. For all three types of knowledge, the RWL condition led to the highest scores, while the listening condition resulted in the lowest scores. All three knowledge types also showed significant differences between input conditions. However, the size of the effect of input condition on the scores varied between the types of vocabulary knowledge types. For meaning recall, a moderate effect was found, while only small effects were found for form recognition and meaning recognition, with the

effect on form recognition being the smallest out of the three. Evidently, input condition affects all types of vocabulary knowledge, but has a larger effect on meaning recall, and less impact on meaning recognition, and less effect still on form recognition. For form

recognition, a significant difference was found only between listening and RWL, possibly due to the small effect size, only the highest scores differed significantly from the lowest scores. For meaning recognition and meaning recall, significant differences were found both between reading and listening, and between listening and RWL. None of the vocabulary knowledge types showed significant differences in scores obtained in the reading and RWL conditions.

4.3 Retention

In order to investigate the retention of the vocabulary knowledge acquired in different input conditions over a period of two weeks, the vocabulary scores on the delayed post-test were compared. Like the immediate test, four scores were assigned for the delayed post-test, form recognition, meaning recognition, meaning recall, and the overall score that combines all three vocabulary knowledge types. The descriptive statistics for the overall vocabulary scores on the delayed post-test can be found in Table 7.

Table 7. Descriptive statistics for the overall vocabulary scores obtained on the delayed

post-test in the reading, listening, RWL conditions.

Input condition Mean Min Max SD

Reading 4.17 2 6 0.99

Listening 4.01 2 5.33 1.00

A Friedman test found a significant difference between the overall vocabulary scores attained in the three different input conditions, χ2(2, 28) = 7.63, p = 0.02, and Kendall’s W found this to be a small effect, W = 0.116. A paired Wilcoxon signed-rank test found a significant difference between the overall scores in the listening and RWL conditions, p = 0.015. There were no significant differences between reading and listening, and reading and RWL. The overall vocabulary scores on the delayed post-test are visualized in Figure 5.

Figure 5. Boxplot showing the overall vocabulary scores on the delayed post-test in the

reading, listening, and RWL conditions.

As can be seen in Figure 5, the overall scores for the three input conditions lie close together, with RWL showing the highest mean scores. While the mean scores of listening (4.01) and reading (4.17) are both lower, a statistically significant difference to the RWL scores was only found for the listening condition.

Table 8 shows the mean, minimum, maximum, and standard deviation for the form recognition scores in each input condition.

Table 8. Descriptive statistics for the form recognition scores obtained on the delayed

post-test in the reading, listening, RWL conditions.

Input condition Mean Min Max SD

Reading 7.55 5 10 1.50

Listening 7.79 4 10 1.59

RWL 8.66 4 10 1.59

A significant difference between the form recognition scores of the different groups was found using a Friedman test, χ2(2, 28) = 10.02, p = 0.007. Kendall’s W showed this to be a small effect, W = 0.152. A paired Wilcoxon signed-rank test found a significant difference between the form recognition scores in the reading and RWL groups. No significant

differences were found between reading and listening, and listening and RWL. The form recognition scores on the delayed post-test are visualized in Figure 6.

Figure 6. Boxplot showing the form recognition scores on the delayed post-test in the

reading, listening, and RWL conditions.

Figure 6 shows the highest form recognition scores for the RWL condition, and the lowest for reading. This is in accordance with the significant difference found between the scores attained in these two input conditions.

The descriptive statistics for the meaning recognition scores on the delayed post-test are shown in Table 9.

Table 9. Descriptive statistics for the meaning recognition scores obtained on the delayed

post-test in the reading, listening, RWL conditions.

Input condition Mean Min Max SD

Reading 4.21 1 5 1.15

Listening 3.69 1 5 1.26

RWL 4.14 2 5 0.88

A Friedman test found significant differences between the meaning recognition scores, χ2(2, 28) = 7.26, p = 0.02. Using Kendall’s W, this was found to be a small effect, W = 0.110. However, the paired Wilcoxon signed-rank test did not find significant differences between any of the groups. Figure 7 shows a visualization of the meaning recognition scores.

Figure 7. Boxplot showing the meaning recognition scores on the delayed post-test in the

As can be seen in Figure 7, the meaning recognition scores were fairly high for every input condition, and while the highest scores were attained in the reading condition, no significant differences were found between any of the conditions.

Table 10 shows the descriptive statistics for the meaning recall tests on the delayed post-test.

Table 10. Descriptive statistics for the meaning recall scores obtained on the delayed

post-test in the reading, listening, RWL conditions.

Input condition Mean Min Max SD

Reading 0.76 0 3 1.06

Listening 0.55 0 3 0.83

RWL 1.24 0 5 1.48

A Friedman test showed a significant difference between the meaning recall scores in the three different input conditions, χ2(2, 28) = 6.35, p = 0.04, and Kendall’s W found this to be a small effect, W = 0.096. However, like the meaning recognition scores, a paired Wilcoxon signed-rank test did not find any significant differences between any of the input conditions. The meaning recall scores on the delayed post-tests are visualized in Figure 8.