The effect of type of subtitles and proficiency on vocabulary acquisition

Subtitles and vocabulary acquisition

The effect of type of subtitles and proficiency on vocabulary acquisition

Mirte Wilhelm S2505088

MA in Applied Linguistics Faculty of Liberal Arts University of Groningen

Supervisor: Dr. H. Loerts Second reader: Prof. dr. C. L. J. de Bot

March 11th, 2018 Word count: 16,502

Contents

Abstract ... 3

Introduction ... 4

1 Theoretical Background ... 5

1.1 Movies as Authentic Input ... 5

1.1.1 Advantages of watching movies for language learning. ... 6

1.1.2 Drawbacks of using movies in the classroom. ... 8

1.2 The Use of Subtitles ... 10

1.2.1 Intralingual subtitles. ... 11

1.2.2 Interlingual subtitles. ... 13

1.2.3 Interlingual subtitles vs. intralingual subtitles. ... 14

1.3 Contradictory Findings ... 16

1.3.1 The role of proficiency. ... 16

1.3.2 Processing subtitles. ... 18

1.3.3 Familiarity. ... 20

1.3.4 Linguistic differences. ... 21

1.4 Statement of Purpose ... 22

2 Method ... 23

2.1 Participants ... 24

2.1.1 Age, gender and location. ... 24

2.1.2 Educational level. ... 25

2.1.3 Linguistic background... 26

2.1.4 Language use. ... 31

2.2 Material ... 33

2.2.1 Questionnaire. ... 33

2.2.2 Dialang. ... 34

2.2.3 Vocabulary test. ... 34

2.2.4 Movie excerpt. ... 38

2.2.5 Subtitles. ... 39

2.2.6 Eye-tracker. ... 40

2.3 Research Design and Procedure ... 41

2.3.1 Research design... 41

2.3.2 Procedures. ... 42

2.4 Analysis... 44

3 Results ... 44

3.1 Description of All Scores ... 44

3.2 Effect of Proficiency on the Learning Effect ... 47

3.3 Content Questions ... 48

3.4 Eye Tracking Data ... 50

3.4.1 The subtitles in total. ... 50

3.4.2 Correlation between the eye tracking data and the learning effect. ... 51

3.4.3 Individual words. ... 52

5. Discussion ... 56

5.1 The Learning Effect ... 56

5.2 The Proficiency ... 57

5.3 Subtitle Type ... 58

5.4 Eye Tracking Data ... 59

5.5 Meaningfulness Results ... 59

6. Conclusion ... 60

6.1 Contextualisation ... 60

6.2 Suggestions and Recommendations for Further Research ... 62

References ... 64

Abstract

This thesis investigates which type of subtitles, interlingual or intralingual, are best for vocabulary acquisition of Dutch learners of Swedish and whether the effectivity of the different type of subtitles depends on the proficiency level of the learners. It is hypothesized that the interlingual subtitles are more suitable for beginners, while interlingual subtitles are the most beneficial for the vocabulary acquisition of more

proficient learners. To examine this, twenty learners of Swedish were examined, half of them with low proficiency and half of them with high proficiency in Swedish. The participants watched a ten minute Swedish-spoken video with either Dutch (interlingual) or Swedish

(intralingual) subtitles, while their eye movements were recorded. During pre- and post-tests they had to write down the translation of words that were cut from the audio of the movie of which a fragment was shown. The post-test also included some comprehension questions and the

question whether the words in the test occurred in the video. A repeated measures ANOVA revealed that, overall, the more proficient participants showed a higher learning effect than the lower leveled participants. The type of subtitles did not have any effect. The eye tracking data showed that the participants looked at the subtitles, but no relation between the time spent and the learning effect could be seen on an individual word level.

Keywords: Subtitles, intralingual subtitles, interlingual subtitles, Swedish, Dutch, second

language acquisition, eye tracker

Introduction

The aim of this thesis is to investigate how beneficial different types of subtitles are for second language vocabulary acquisition. Watching movies in the target language is increasingly popular in second language education. This is understandable, as movies provide a glimpse of the target culture and provide the learner with authentic use of the language. Furthermore, movies are very engaging, which makes watching a movie in the classroom an attractive activity for the learners.

The second language teacher can decide to show the movie in the target language with subtitles to support the learners. There are different possibilities to choose from. The subtitles can show the translation of the audio into another language, usually the native language of the

learner. These are called interlingual subtitles. Subtitles that are written in the same language as

the audio of the movie are called intralingual subtitles. Of course, the teacher can also choose to

not turn on any subtitles when watching a movie in the target language. When choosing, it is

important to know the effect that the different types of subtitles have and which type of subtitles

is the most suitable in which stage of the acquisition process.

1 Theoretical Background

In this section, the prior research that has been carried out on the topic of the use of subtitles with the aim of language acquisition will be discussed. Firstly, the focus is on the use of authentic input and language learning, more specifically the benefits and drawbacks of watching movies in the second language classroom. Secondly, the use of subtitles and effects of different types of subtitles on language acquisition in prior studies are presented. Lastly, light will be shed on some aspects that could explain these findings and the statement of purpose is also part of this section.

1.1 Movies as Authentic Input

Subtitles are used to support the watching of movies, and showing movies in a foreign language to learners is a way of providing authentic material. Viewing movies in a foreign language can be beneficial for language learning. In particular, movies can support the

vocabulary knowledge of language learners. Additionally, movies are engaging for the learners, which makes them both enjoyable and instructive.

The use of authentic input, such as foreign language movies, is considered to be important in second language learning. This is especially true within the context of Communicative

Language Teaching (CLT) and Content and Language Integrated Learning (CLIL) (Pinner, 2013, p. 53). According to adherents of usage based language teaching, learning a second language is similar to learning a native language (Krashen, 1982 as cited in Bot, Lowie, & Verspoor, 2005, p.

79). To imitate the circumstances that make native language learners successful, it is important for second language learners to get as much exposure to the target language as possible.

The importance of exposure is also supported by Krashen’s input hypothesis (1985, as

cited in De Bot, Lowie, & Verspoor, 2005, p. 36). According to this theory, language learners

acquire a language through exposure to the natural language, as opposed to studying grammar. A requirement is that the second language input is on a slightly higher level than the present level of the student. This is illustrated as a “i+1”, i being the current stage of development and the 1 the extra difficulty that is needed for further language acquisition. The suitable second language input is called ‘comprehensible’, as the language should still be understandable for the learner (1985, as cited in De Bot, Lowie, & Verspoor, 2005, p. 36).

Thus, according to Krashen’s theory, comprehensible input should be used to enable the language learner to acquire a second language as successfully as possible. Authentic materials can be used to realize this desired exposure to comprehensible input during the language learning process. Pinner defines authentic material as materials that were not created for the sake of language learning or description of a language, but rather created with the goal of

communication (2013, p. 61). This definition includes movies that are produced for native speakers of the language that learners want to acquire.

1.1.1 Advantages of watching movies for language learning. In line with Krashen’s theory, that is explained above, many advantages for language learning are accounted to

authentic input. An account of these advantages is given in the following section. The focus is on the use of movies in the target language as authentic input.

Experts on language teaching agree that communication skills of learners increase when

authentic input is used (Akbari, Razavi, & Azam, 2016, p. 107). Another reason to use authentic

input is to accustom the learner with the target language as they may encounter it in real life

(Pinner, 2013, p. 59). And although Gilmore (2007) points out that the research on the topic is

limited, some studies have shown that watching movies is motivating for language learners

(Webb & Rodgers, 2009, p. 1). The wide variances of genres and topics guarantees that learners can find a movie that fits their taste. Consequently, exciting stories will make the students emotionally involved, which ensures that their attention will be on the screen (King, 2002, p.

515). Besides being entertaining, the story of the movie provides context to the utterances in the movie. This encourages incidental learning (Nakamura, 2007, p. 126). Turning now to consider the occurrence of incidental learning and the context in movies in more detail.

Incidental learning is learning that takes place unintentionally. It occurs subconsciously, when focusing on the meaning of an utterance. An example is the acquirement of vocabulary while reading. When a learner reads a text and repeatedly encounters an unknown word, the learner can guess the meaning of the word using the context, and consequently acquire the new word without consulting a dictionary (Bot et al., 2005; Ponniah, 2011). Incidental learning also occurs when the input is presented through audiovisual media. For example, a study investigating the effect of longitudinal media exposure on 374 Flemish students in primary school showed that Flemish learners of English that self-reported to be more exposed to English media performed better on oral translation tests (Kuppens, 2010).

What might enhance incidental vocabulary learning in the case of movies is the fact that movies are multisensory or intersemiotic. A movie excerpt provides the learner with both aural and visual input at once providing both contextual and visual cues. Examples of contextual cues are intonation, sounds in the background and the quality of voice. These cues that are given through the audio channel can be either verbal or non-verbal (such as background noises).

Examples of visual cues that can provide more information about the content are the body

language of the characters, their facial expressions and physical actions, the costumes, the setting

of the scene and even visible objects that the characters are referring to (Baltova, 1999, p. 34).

All these cues contribute to incidental learning through video materials (Webb & Rodgers, 2009, p. 6).

1.1.2 Drawbacks of using movies in the classroom. Having demonstrated that the use of movies as authentic material to enhance language acquisition has its benefits, the next section will address some drawbacks. Firstly, the entertainment value of movies has lead teachers to consider the use of movies as something that only has a place in the learners’ free time. Although using movies for language teaching is becoming more popular, the pedagogical value of movies is not always acknowledged. Showing movies is not included in the standard curriculum of language teachers, so it is difficult to fit the activity in the existing schedule (King, 2002, p. 511).

As a result, watching a movie in the target language is used sparingly as a teaching method and it is often regarded as a “Friday afternoon treat” (MacKnight, 1983, p. 10). Usually, the movies are shown separately and without additional support. This approach disregards the possible value movies can have for the acquisition process (MacKnight, 1983).

Another disadvantage is that watching a movie in a foreign language can be a challenge

even for more proficient second language learners. Because the language in movies is directed at

native speakers, the dialogues might be hard to follow for someone who does not regularly hear

the language in daily life. For example, the voices in the movie can give rise to comprehension

problems. This happens when people in the movie mumble or have regional accents. Learners of

English as a second language that have learned British English in school might for instance have

difficulty comprehending the speech in American movies. Of course, a language learner might

not be able to process rapid speech and have difficulty distinguishing different words in a

sentence.

Furthermore, unknown vocabulary items that are used in movie dialogues may prevent learners from fully comprehending the content. Oftentimes technical or specialized terminology is used that goes beyond the vocabulary level of second language learners. Similarly, different registers are used that the learner might not be familiar with. Informal use of the language includes slang that is unlikely to be covered in the school curriculum. Use of idioms that the viewer is unfamiliar with can lead to confusion as well (King, 2002). Research on this topic has established that learners have to understand 90% to 99% of the words that are used in a movie to be able to follow it without subtitles, so the vocabulary of the movie can only deviate a little from the lexicon of the learner (Webb & Rodgers, 2009). This is consistent with Krashen’s Comprehensible Input theory, which was described earlier.

With respect to the lexicon of the learner, a study by Webb and Rogers (2009) suggests a vocabulary size of approximately 3000 of the most frequent word families (excluding proper nouns and marginal words) is sufficient to be able to follow movies in the L2. This vocabulary has been found to cover on average 95% of the vocabulary used in English spoken movies (Webb & Rodgers, 2009, p. 18). Although the description of the CEFR levels do not specify the vocabulary size that is required to reach the different levels, Milton (2010) has investigated which CEFR level roughly correlates with which vocabulary breadth. According to his study, knowledge of the most frequent 3000 words is needed to reach the B1 and B2 levels, indicating that L2 movies without subtitles are understandable for learners that have passed at least level A2 (Milton, 2010, p. 226).

Even when the learner can decipher all the words and their meaning, cultural differences

can prevent full understanding of the movie. The language learner can misinterpret meanings due

to a lack of understanding of the pragmatics of a different cultural environment. Cultural-specific

expressions or concepts are foreign and unintelligible to the language learner. Other aspects of a movie that people that are unfamiliar with the culture might miss are cultural references that assume knowledge of specific events or people that are not widely known in different cultures.

That and a possibly different type of humor may cause the jokes in a movie to not have the intended effect on the foreign viewer (King, 2002). As a result, the viewer will always have a different experience from a native speaker, even when watching the same movie, because they do not belong to the target audience (Baltova, 1999).

1.2 The Use of Subtitles

Subtitles can be used to overcome these linguistic and cultural hurdles and make a foreign

movie more accessible for language learners. In subtitled movies the spoken text is displayed on

the screen. There are different types of subtitles. The subtitles can be in the same language as the

audio, as a transcript of the spoken language. Those type of subtitles are called intralingual

subtitles, captioning or karaoke-subtitles. In this paper they will be referred to as intralingual

subtitles. Interlingual subtitles differ from intralingual subtitles as they provide a translation of

the foreign language audio. Thus, the subtitles are displayed in a different language from the

audio. This is also the case for reversed subtitles. However, with reversed subtitles the languages

are switched, and the audio track is played in the native language, while the subtitles are in a

language that is foreign to the watcher. All of these three subtitle types can help language

learners understanding audiovisual input and enhancing their language proficiency. Because the

interlingual and intralingual subtitles are included in the experiment of the present study, these

subtitle types and their benefits will be discussed in more detail.

1.2.1 Intralingual subtitles. Intralingual subtitles were originally designed to make movies accessible to the hearing impaired. By providing the visual transcription of the spoken language, processing the audio is no longer a prerequisite to understand what is going on.

Fortuitously, exactly the combination of the audio and the written subtitles make this type of subtitles not only suitable for the hearing impaired, but also for language learners. They provide a visualization of the audio. The language learners, that do not belong to the target audience of the audiovisual material and may find comprehension of the audio challenging, are thereby supported by the intralingual subtitles (Danan, 2004).

The following is a more detailed description of how intralingual subtitles can help students to overcome some of the difficulties that arise when watching a movie in a foreign language. For instance, the presence of these subtitles facilitates the familiarization with unknown ways of pronouncing sounds (Vanderplank, 2016a). This makes deviations from the standard pronunciation in the audio, such as the occurrence of different regional accents, less problematic for the student’s understanding of the input. Unclear audio due to shortcomings in quality or background noises is also less troublesome when intralingual subtitles are used to show a transcription of the audio (Winke, Gass, & Sydorenko, 2010).

What’s more, the interspacing in the written text provides the viewer with information about word boundaries that is not present in the speech. Qualitative research confirms that

intralingual subtitles facilitate indicating word boundaries while listening to the foreign language, which makes it easier to distinguish words (Winke et al., 2010). One of the challenges of

listening comprehension is to discern words in the speech stream (Graham, 2006). The

identification of word boundaries supports form-to-meaning-mapping (Winke et al., 2010). A

study by Lees (2014) that investigated intermediate Japanese learners of English confirms that

watching a video with intralingual subtitles resulted in better results on vocabulary perception than both interlingual subtitles and no subtitles (Lees, 2014).

Now that the advantages of watching movies with intralingual subtitles are explained, the experimental evidence for the effect of intralingual subtitles on language acquisition will be discussed, starting with a study done by Winke, Glass and Sydorenko. The result of this study shows that watching videos with intralingual subtitles is more effective for language learners than watching videos without subtitles (Winke et al., 2010). Twenty-six American students learning vastly different languages (Arabic, Chinese. Spanish and Russian), were shown three short videos in their target language. Some videos included intralingual subtitles and other videos were not subtitled. The tests were presented both aurally and written. Watching intralingual videos resulted in significantly better results for novel vocabulary recognition exercises than watching a video without subtitles. The comprehension was also better for the students that watched the videos with intralingual subtitles compared to the control group.

A meta analyses carried out by Montero Perez, Van den Noortgate and Desmet confirms this positive effect (2013). They looked at the effectiveness of eighteen studies concerning the effect of intralingual subtitling on listening comprehension and vocabulary learning. The findings show an overall large effect of watching video material with intralingual subtitles on both

comprehension and the vocabulary recognition compared to control groups that watched videos without subtitling. Furthermore, no confirmation bias on this subject was found in the meta study, as the results of the published and unpublished studies did not differ significantly. Overall, it can be concluded that intralingual subtitles contribute to listening comprehension and

vocabulary learning.

However, a prerequisite for the suitability of intralingual subtitles for language learners is an accurate transition from the audio to text. Due to the rapid speech of the audio and the

limitations of time and space that apply to subtitles, some adjustments to the original text must be made to fit into the subtitles. Despite this, the sentences should stay as close to the original soundtrack as possible, and should not be paraphrased for the subtitles to be effective in terms of language acquisition. A large difference between the spoken text and the provided subtitles does not support understanding, but creates semantic confusion instead (Caimi, 2006). Another type of subtitles that does provide the translation to the mother tongue is interlingual use of subtitles, which will be explored in the next section.

1.2.2 Interlingual subtitles. Interlingual subtitles are a popular way to make foreign television shows and movies available for broadcasting in small countries. It is a cheaper alternative to dubbing, as dubbing is 10 to 15 times as expensive (Koolstra, Peeters, & Spinhof, 2002). Countries in Europe where subtitles prevail are marked in yellow on the map below.

Those countries include for example the Scandinavian countries and the Netherlands. In other

countries, such as Germany and France, dubbing is the usual practice for televised programs and

movies (Media Consulting Group, 2011, p. 8). Although these countries can traditionally be

considered as dubbing countries, subtitles have become increasingly widespread even there since

the existence of the internet. To assume that people in dubbing countries do not watch media

with interlingual subtitles is therefore outdated (Perego, 2016).

Figure 1 Map of language transfer practices (subtitling, dubbing, voice-over or originals) in Europe (European Commission, 2011, p. 9)

With respect to the benefits of interlingual subtitles for language acquisition, most research has been done in the 90s and the main consensus is that the interlingual subtitles could be helpful for vocabulary learning (Danan, 2004). A more recent study by Ina (2014) supports this. In this study Greek children with no prior knowledge of Italian were shown a television show in three conditions: Italian soundtrack with Greek subtitles (interlingual), Italian soundtrack without subtitles and Greek soundtrack without subtitles (control). The children that had watched the video with interlingual subtitles had acquired the most Italian vocabulary and scored better on word recognition tests than participants that had watched videos with the other conditions (Ina, 2014).

1.2.3 Interlingual subtitles vs. intralingual subtitles. Although the focus of research on

subtitles in Second Language Acquisition (SLA) has been on intralingual subtitles in the 1990s,

more recent research has shown interest in interlingual subtitles as well. As both videos with

intralingual and interlingual subtitles have been found to enhance comprehension and vocabulary

learning compared to videos without subtitles, the question arose which type of subtitles is the

most effective for language learners. To find the answer to this question, some studies have been carried out to compare the effect of intralingual subtitles and interlingual subtitles on

comprehension and vocabulary acquisition (e.g., Matielo, D’Ely, & Baretta, 2015, p. 179).

Some findings suggest that interlingual subtitles are more effective for vocabulary learning than intralingual subtitles. The first one is a study by Bianchi and Ciabattoni (2008), that investigated the effect of the different types of subtitles on the comprehension and vocabulary results of adult Italian learners of English. The participants had a wide range of proficiencies.

Overall, both the content comprehension and the vocabulary comprehension turned out to benefit most from the interlingual subtitles. Another study in which is found that interlingual subtitles are more effective than intralingual subtitles is a recent study investigating vocabulary learning of Iranian speakers of English (Mardani & Najmabadi, 2016, p. 61). Relevant to note here is that the reading direction of Persian is right-to-left, contrary to English. This might affect the Iranian participants’ reading capability of the intralingual subtitles compared to the interlingual subtitles.

This aspect will be discussed later on.

As indicated previously, there are studies that found that interlingual subtitles are the most effective for language acquisition. Yet, there are studies that contradict these findings and found that intralingual subtitles are more beneficial for language learners than interlingual subtitles.

Peters and his colleagues, for example, found that, although both intralingual and interlingual

subtitles enhance vocabulary learning, learners benefit more from subtitles in their native language

(Peters, Heynen, & Puimège, 2016). Additionally, a study by Lees found the perception and

noticing of L2 vocabulary to be better for participants that had watched a video with intralingual

subtitles (Lees, 2014). Another study that investigated the ability to repeat certain words from

videos found that interlingual subtitles harmed the performance due to lexical interference, while the intralingual subtitles assisted speech learning (Mitterer & McQueen, 2009).

1.3 Contradictory Findings

These contradicting results might be caused by the large differences between studies that have focused on different types of subtitles. The participants in the studies varied for example in age, familiarity with subtitles and proficiency in the L2. The language pairs which are in focus and their writing systems also differ vastly within the research. These aspects will be further discussed in the following section. The variation between the different studies that focus on subtitles might contribute to a more universal view on the effect in subtitles, but it also raises questions about to what degree the findings of studies can be compared. The proficiency of the participants in particular can have a determining influence on the results. As the suitability of subtitles for different stages of the foreign language development are the subject of this study, the role of proficiency will be scrutinized next.

1.3.1 The role of proficiency. It is often assumed that interlingual subtitles are most suitable for beginning learners and intralingual subtitles are most suitable for more proficient learners (Matielo et al., 2015, p. 172). Vanderplank argues that learners need a certain

proficiency in reading the L2 before they can benefit from intralingual subtitles (Vanderplank,

2010). Not only must they have some reading comprehension in the L2 to be able to understand

the subtitles, the limited amount of time that the subtitles are visible make a high reading speed

in the L2 also a necessity (Vanderplank, 2016b). Similarly, Danan stresses that the level of the

language uttered in the input and subtitles should not deviate too much from the current level of

the learners (Danan, 2004). That idea is in line with Krashen’s Comprehensible Input theory, that

states that input is only helpful when it is slightly above the current capability of the student but still within reach (Krashen, 1985).

Most of the studies discussed above confirm the statement that interlingual subtitles are more suitable for beginners and intralingual subtitles are more suitable for more advanced learners.

Overall, the Italian adult learners of English in Bianchi and Ciabattoni’s study benefited the most from interlingual subtitles. When proficiency is taken into consideration, the beginners benefited the most from interlingual subtitles, while intralingual subtitles proved to be more helpful for intermediate and advanced learners (Bianchi & Ciabattoni, 2008). The intermediate learners in Lees’ study were also supported best by intralingual subtitles (Lees, 2014). Furthermore, the intralingual type was also found the most effective for speech perception of the advanced Dutch learners of English (Mitterer & McQueen, 2009).

Nevertheless, the studies do not unanimously fit into this view. Peters, Heynen and Puimège found that not only intermediate learners, but also beginners benefited more from intralingual subtitles than from interlingual subtitles in word knowledge exercises (Peters et al., 2016). The long term results in the Bianchi and Ciabattoni study do not agree with the main consensus either. When the participants were tested again, one week after they watched the video, the beginners appeared to have benefited the most from the intralingual subtitles, while the intermediate advanced learners had benefited the most from the subtitles in their native language (Bianchi & Ciabattoni, 2008). This is interesting, and indicates that longitudinal effects can be different from short-term effects and are worth examining.

To sum up, which type of subtitles are the best suited in which stage of the developmental

process remains unclear. However, both the theory and the experimental research show a clear

trend: for language learning purposes, the interlingual subtitles are supposedly more appropriate for learners that are in the beginning stage of second language development, while the intralingual subtitles tend to be more suitable for more advanced learners. The studies that support this used comprehension tests and vocabulary tests, and short-term results that were collected immediately after showing the (subtitled) video.

1.3.2 Processing subtitles. Another vexed issue, apart from the role of proficiency, is the degree to which subtitles distract viewers from both the foreign audio and the video material. For example, some American students that were in their second year of language classes (Spanish and Chinese, respectively), felt that they had less attention for the actual video when they were reading the subtitles (Winke et al., 2010, p. 78). That is not the only study in which the

participants indicated that they found the subtitles distracting; it is an often mentioned complaint, regardless of the type of subtitles.

A reason for this reaction might be that the participants experience processing difficulties.

Subtitles add another channel to the semiotic channels; besides the picture in the visual channel and the dialogue, music and sound effects in the audio channel, there is a verbal visual channel (Lertola, 2015). This multi-semiotic nature of videos with subtitles requires learners to listen, read and look at the visual picture at the same time (Caimi, 2006). This requires processing skills, which might cause people who have difficulty internalizing the information from all the channels at once to feel distracted.

Likewise, an often mentioned disadvantage for the learning benefits of interlingual

subtitles is that viewers can ignore the L2 audio because they only have to read the translation to

be able to comprehend the conveyed meaning. That way, they would not pay attention to the L2

input and consequently not benefit from it. The findings that are described in a report by the

European Commission concerning the language competence of Europeans confirms this theory.

The CEFR levels of people that watched English media without subtitles were compared to the level of those who self-reported to watch media with interlingual subtitles. Watching media with interlingual subtitles turned out to have little impact on the proficiency level (Araújo & Dinis Da Costa, 2013).

On the other hand, there are findings that suggest that reading subtitles does not prevent viewers from processing the soundtrack of the video (Danan, 2004). Vanderplank also argues that people who watch subtitled videos do not totally block out the audio, but try to match sound and text as they watch (Vanderplank, 1988). Similarly, research has shown that simultaneously processing subtitles, sound and image at the same time is perfectly possible (Perego, 2016).

Everybody automatically reads subtitles (d’Ydewalle, 2002). This automatic reading of subtitles has even been observed in Americans that were watching a movie with Dutch subtitles, which means that it even occurs when people are confronted with a foreign language they are unfamiliar with (Bisson, Van Heuven, Conklin, & Tunney, 2014). Generally, children can read fast enough to be able to follow L1 subtitles when they are about ten years old (Vanderplank, 2016b, p. 239).

Processing subtitles takes effort, but the processing ability can also be trained. The average fixation time on intralingual subtitles has been observed to be larger than usual, which could be expected, as the participants were reading and listening at the same time, with the added linguistic difficulty, which increases the challenge (Bisson et al., 2014). This suggests that

processing subtitles is difficult. Training can however increase the chunking ability of learners.

By watching videos with intralingual subtitles people are trained to process more text and audio

at once. This was noted after a few hours of watching, so this may not be an immediate effect

(Vanderplank, 2016a, p. 52). Nevertheless, this raises the question which role familiarity to subtitles plays. That will be the subject of the next section.

1.3.3 Familiarity. A cause for the fact that many learners experience the subtitles as distracting could be their lack of familiarity with them. This would be the case for the American students mentioned earlier. Processing the sound, video and subtitles at once takes effort, and people that are not familiar with subtitles might experience subtitles as distracting while they try to focus on everything at once. Once they get used to subtitles they develop strategies that help them deal with the multisensory input (Danan, 2004). The findings of a cross-national study carried out by Perego suggest that the lack of familiarity with subtitles does indeed affect the attitude towards subtitles, yet the subtitles proved to be effective for all language learners regardless of familiarity (Perego, 2016).

Another reason people find subtitles distractive could be a too large gap between the learners’ language proficiency and the level of the language that is used in the video. For

example, the first-year students that participated in Taylor’s (2005) study concerning intralingual subtitles found the subtitles more distracting than third-year students. As Vanderplank argues, this could be due to the low proficiency of first-year students (Vanderplank, 2010). In another study, he mentions that the reason that Arabic learners of English found subtitles distracting was their relatively slow reading speed. Although they could follow the spoken audio without much difficulty, they could not keep up with the subtitles. The subtitles made them more aware of their slow reading speed in English. This could have been caused by a lack of proficiency

(Vanderplank, 1988). However, the difference in writing direction may also be the cause of the

slow reading speed of the participants. Thus, the experienced distraction was caused either by the

proficiency of the learners or by the linguistic differences between the languages of focus. The

effect of differences between languages will be considered in the following section, specifically light is shed on the orthographic differences that are relevant regarding subtitles.

1.3.4 Linguistic differences. This study focuses on Dutch learners of Swedish. Dutch people are used to subtitled television. The subtitled television they are most used to consists of English spoken programs with interlingual subtitles. In her study Bos (2014) has tested the effect of interlingual and intralingual subtitles on the vocabulary acquisition of Dutch learners of English. The focus on Swedish distinguishes the two studies in several ways.

Firstly, Swedish is a much smaller language than English. This study will focus on Swedish spoken footage with subtitles in Dutch and Swedish for Dutch learners of Swedish. Swedish is a small language, so exposure to English is more frequent for people in the Netherlands than exposure to Swedish. In this regard, Swedish is similar to Danish. A former study has revealed that watching a 10 minute long, interlingually subtitled movie of Danish increased the vocabulary knowledge of Dutch participants that did not have any former knowledge of Danish (d’Ydewalle

& Van de Poel, 1999).

Besides the difference in the amount of input, the difference between English and Swedish

that might play a role lies in the orthographic depth of the languages. In a shallow writing system

there is a clear correspondence of the letters to the phonemes: each letter consistently represents a

sound. In writing systems with deep orthographies on the other hand there are more exceptions,

and irregularities, which make the writing system further removed from the spoken language. The

orthography of Swedish is more shallow than the deep orthography of English (Seymour, Aro,

Erskine, Wimmer, & Schneider, 2003, p. 147). This could make Swedish intralingual subtitles

easier to follow for Dutch learners than English intralingual subtitles, because the text can more easily be related to the spoken language that is played simultaneously.

Both English, Swedish and Dutch use the Latin alphabet. The reading direction is from left to right. The alphabet used by in English and Dutch is identical and consists of 26 letters. The Swedish alphabet on the other hand includes three extra letters: the å, ä and ö (Institutet för språk och folkminnen, 2014). These letters are new to Dutch learners of Swedish, although learners that have been in contact with German might already be familiar with the ä and ö, that are also used in German. These three extra letters might make reading intralingual subtitles in Swedish slightly more challenging for Dutch learners of Swedish.

In terms of orthographical differences, Winke, Gas and Sydorenko argue that learners of languages that are orthographically different rely more on listening. These learners that watch a movie with intralingual subtitles have more difficulty following the text that is presented in an unknown orthography. This was noted in regard to native speakers of English learning languages with similar orthographies (Spanish and Russian) versus languages with a vastly different orthographic system (Chinese and Arabic) (Winke et al., 2010).

Compared to those languages, the differences between the English and Swedish orthographies is only small. The effect of the different orthography of Swedish compared to English would probably be less noticeable, as they both use the Latin alphabet and the reading direction does not differ between the languages.

1.4 Statement of Purpose

In conclusion, research shows that both interlingual and intralingual subtitles support

comprehension and vocabulary acquisition of viewers. The effects of different types of subtitles

on the second language acquisition remains inconclusive, especially when the proficiency level of language learners has been taken into consideration. Furthermore, most of the conducted studies are concerned with learners of English, a language that language learners are likely exposed to in their daily lives through English spoken media. Other studies focus on languages such as Arabic and Chinese, which have a vastly different orthographic system than English. Swedish on the other hand is a smaller language which has an orthography that is more similar to Dutch. By using Swedish in the experiment, the possible effects of differences in orthography between the languages can be ruled out, which enables a clear focus on the effect of the learners’ proficiency.

The aim of this study is to examine which type of subtitles, interlingual or intralingual, are best for vocabulary acquisition of Dutch learners of Swedish. The main aim is to investigate whether the effectivity of the different type of subtitles depends on the proficiency level of the learners. Most of the literature suggests that the effectiveness of different types of subtitles for language learning is indeed dependent on the stage in the linguistic development of the learner.

Particularly, interlingual subtitles would be more beneficial for beginners and intralingual subtitles would be more beneficial for advanced learners.

2 Method

To find an answer to the research questions, a quantitative experiment was created in

which participants with different levels of proficiency in Swedish watched a Swedish spoken

movie excerpt in two different conditions: one with Dutch subtitles (interlingual), and one with

Swedish subtitles (intralingual). In this method section, the participants that agreed to engage in

the experiment are described, just as the materials that were used to conduct the experiment and

the analysis that was used to detract result from the obtained data. Firstly, the information about

the participants will be presented.

2.1 Participants

Twenty learners of Swedish participated in the experiment of the present study. They had followed or were following Swedish language courses, so they met the requirement of being learners of Swedish. The participants were either students at the University of Groningen or registered privately at the Scandinavisch Vertaal- en Informatiebureau Nederland (SVIN or Scandinavian Translation- and Information office in The Netherlands).

2.1.1 Age, gender and location. Only 30% (six) of the participants were male. All participants were adults, the youngest being 19 years old and the oldest 59. This was caused by the fact that there were not enough students between 18 and 25 who studied Swedish at the University of Groningen that could participate in the experiment. The range was quite large (40), but the median of the age was 23 and a half and the mode was 22, so most of the participants were around twenty years old. For the descriptive statistics see Table 1 below.

Table 1 Descriptive statistics of the age of the twenty participants that attended the experiment

Average 32,10

Median 23,5

Mode 22

SD 13,57591

Minimum 19

Maximum 59

Turning now to the participants’ descent. The majority of the participants was Dutch,

only four of them were born outside of the Netherlands. Two of them originated from the United

Kingdom, one from Germany and one from Finland. All of the non-Dutch participants were

students of Swedish at the University of Groningen. All these participants were exposed to the

excerpt with intralingual subtitles, because they would not be able to understand the Dutch

subtitles as well as the native speakers of Dutch. Three of the four non-Dutch participants had a high proficiency in Swedish.

Figure 2 The province of birthplace of the Dutch participants

As can be seen in Figure 2 above, most of the Dutch participants came from the East of the Netherlands, most notably the provinces Groningen, Drenthe and Gelderland. This is not surprising, because the testing took place in Groningen and the participants were enrolled in classes that were taught in Groningen. The following is a brief description of the educational background of the participants.

2.1.2 Educational level. One of the questions in the questionnaire was the highest completed level of education. Half of the participants answered that high school was the highest educational level that they had completed. Nine of these participants were currently enrolled in a Bachelor program at the University of Groningen. Furthermore, five participants had completed higher education at University level (WO), and three had completed an associate degree (HBO).

Province of birthplace

Drenthe Friesland Gelderland Groningen North Holland Overijssel South Holland

Two of the participants had completed a vocational degree (MBO). As such, the educational background of the participants was rather varied.

2.1.3 Linguistic background. Next on the questionnaire were some questions about the linguistic background of the participants and their language use. First, the linguistic background of the participants will be described. The Dutch participants have Dutch as a native language.

One participant learned German and English in addition to Dutch before going to school. The participants from the United Kingdom are native speakers of English, the participants from Germany and Finland are native speakers of German and Finnish respectively. The German participant learned English before going to school.

Most of the people (twelve) that participated in the study have never lived outside of the Netherlands. Apart from the four international students, four participants lived abroad for a period of time. Someone lived in Germany for halve a year and another student lived in England for ten months. What is most relevant for this study is that two of the Dutch participants have completed a semester of their studies in Sweden, in Lund and Uppsala respectively.

All participants have learned other languages than their native language at school and

otherwise. The most frequently mentioned as additional languages besides Swedish were

English, German and French. Figure 3 below shows the occurrence of the different foreign

languages known by the participants. The classical languages are Latin and ancient Greek, that

are taught at some high schools in the Netherlands. For these languages, only reading and

translating are trained, as they have gone extinct.

Figure 3 The foreign languages that the participant report to have learned either in school or otherwise (the classical languages only passively)

The participants were asked to rate their proficiency for every first and second language that they mentioned as known foreign languages. They graded every skill (speaking, listening, writing and reading) on a scale from 1 to 5, 1 being low proficiency and 5 high. All eighteen participants that did not have English as a native language indicated that, after their mother tongue, they were most fluent in English. German and French were the follow-ups, as can be seen by the mean scores that were given for each skill in these languages, displayed in figure 4.

Foreign languages

English French German Classical languages Spanish Dutch Frysian

Figure 4 The Mean self -assessed proficiency level of the participants in English, German and French on a scale from 1 to 5 for the skills speaking, listening, writing and reading

For this study, the level of Swedish of the participants is the most relevant. The average self-assessed score for the different skills in Swedish can be seen in figure 5 below. The

participants estimate their reading and writing skills to be the best, followed by speaking and listening. That listening is considered to be the weakest skill by participants is surprising, as this was not the case for English, German or French (see figure 4). The average grade the participants give themselves for reading in Swedish is 2.9 out of 5, while the average grade for listening is a 1.6. Thus, these participants would on average be more confident following Swedish subtitles than listening to Swedish audio.

0 0,5 1 1,5 2 2,5 3 3,5 4 4,5 5

Speaking Listening Writing Reading

Mean rating of the three most common foreign languages

English German French

Figure 5 The mean self-assessment per skill (speaking, listening, writing, reading) as graded by participants on a scale from 1 to 5

In a subsequent question, the participants were asked to rate their overall proficiency on a scale from one to five. The result can be deducted from Figure 6. The mean rating was 2.2, which is rather low. No one rated their own proficiency with a five, but many participants judged their proficiency with a one. Thus, although the participants seem to be quite confident in speaking and writing, their average estimation of the overall skills in Swedish revealed to be low in comparison.

0 0,5 1 1,5 2 2,5 3 3,5

Speaking Listening Writing reading

Mean rating self-assessment of Swedish proficiency

Swedish proficiency self-assessment

Figure 6 The self -assessed Swedish proficiency level of the participants

The listening and reading proficiency of the participants was also tested using the DIALANG test (“DIALANG [Software].,” 2006), of which the details will be provided in the material section (chapter 2.2). The approximate level of each skill was then revealed through the Common European Framework of Reference (Council of Europe, 2011). The results of the test and the number of occurrence for each combination of listening and reading scores can be seen in table 2 below. Although the participants graded their own listening skills considerably lower than their reading skills, the reading scores did not turn out to be much higher according to the

outcomes of the Dialang test. Often the reading and listening skills were on the same CEFR scale. In four cases the result of the reading test was higher than the result of the listening test, and for the same number of participants the result of the listening test was higher.

0 1 2 3 4 5 6 7 8

1 2 3 4 5

N u m b er o f p ar ticip an ts

Rating on a scale from 1 to 5

Self-judgement Swedish proficiency

Table 2 Swedish proficiency level of the participants in listening and reading on a CEFR scale according to Dialang, the occurrence of each combination of scores and he overall proficiency that corresponds to it. The asterisks mark a missing value.

Listening Reading Occurrence Overall proficiency

A1 * 2 Low

A1 A1 1 Low

A1 B1 1 Low

A2 A1 1 Low

A2 A2 3 Low

A2 B1 2 Low

B1 * 1 High

B1 B1 1 High

B1 B2 1 High

B2 B1 2 High

B2 B2 2 High

C1 C1 2 High

C2 C1 1 High

Because the vocabulary of people with B1 level is enough to be able to understand movies in the target language, the participants with level B1 or higher for both listening and reading were categorized as more proficient (Webb & Rodgers, 2009). Half of the participants were highly proficient according to this system, so both the group with lower proficient participants and the group with more advanced learners consisted of ten people.

2.1.4 Language use. The answers of the participants on the questionnaire questions about

language use will be presented in this section. All Dutch participants speak Dutch on a daily

basis, and the international students speak English daily. Figure 7 below shows how often the participants indicate to speak Swedish. The most common answers are weekly and hardly ever.

Figure 7 Histogram showing the frequency with which the participants indicate to speak Swedish

The participants speak little to no Swedish in daily life. They estimate that the percentage of interactions with family members that happens in Swedish is at most 1%. Furthermore, 16 of the participants do not talk Swedish with friends, of the others there is only one person that gives a higher percentage than three (20%). One person speaks Swedish with colleagues one-tenth percent of the time. They speak Swedish 0-5% of the time when not at home, for only one person this is 30%. Some students of Swedish answered that they speak Swedish during their university classes, which explains that some participants state that they speak Swedish weekly.

Although only a minimum amount of communication happens in Swedish, the percentage of Swedish input through media and reading is somewhat higher. The majority of the participants state that less than one-tenth of their media exposure is Swedish, but for two of the participants

0 1 2 3 4 5 6 7 8 9 10

Hardly ever A few times a year Monthly Weekly Daily

N u m b er o f p eo p le

The frequency with which the participants speak Swedish

Swedish media are more than half of their media exposure. The estimations of the participants concerning the percentage of Swedish input can be seen in Figure 8 below.

Figure 8 A histogram which shows how many participants indicate what percentage of their total media consumption (newspapers, movies, etc.) is Swedish, ranging from 0-10% to 60-70% of their total media consumption

2.2 Material

In this section, the materials that are used to conduct the experiment of the present study are described in detail. The materials include a questionnaire for the participants, a pre-test and a post-test, a proficiency test, a movie excerpt with subtitles, and an eye tracker. The description of the questionnaire will be given first.

2.2.1 Questionnaire. An online questionnaire was used to obtain information about the participants. The questionnaire contained questions about personal details of the participants, their linguistic abilities and their language use. The questionnaire was based on the questionnaire that was used by Bos (2014). There were both a Dutch and an English version of the

0 2 4 6 8 10 12

0-10%

10-20%

20-30%

30-40%

40-50%

50-60%

60-70%

Number of participants

Perc en ta ge o f in p u t

Percentage of Swedish input

Reading Media

questionnaire to cater to both the Dutch and the non-Dutch participants. The links to the questionnaires are provided here:

https://drive.google.com/open?id=1phVNsSzksDEpRhBaFNUN3l7ZBGS0mBRI1gwoshPnyFg https://drive.google.com/open?id=1lmXzBe3fdV4vnPCNrCOSeB9miMqD3yipIzym3kyxPok

2.2.2 Dialang. To be able to compare the performances of participants of varying proficiency levels, the listening and reading skills of the participants were tested. For this purpose, the Dialang test was used. Dialang is an online diagnostic language assessment system that offers language tests for Swedish, amongst other languages. The software is designed by the University of Lancaster, and each language skill (speaking, writing, listening and reading) can be tested separately (“Dialang,” n.d. ) and can be reached through this link:

https://dialangweb.lancaster.ac.uk/. For this study, the listening skills and writing skills were relevant, as the study focuses on the passive knowledge acquired by listening to audio and reading subtitles. Dialang used the Common European Framework of Reference to provide information about the level of proficiency of the participants, which are shown after the test is completed. Consequently, there were two scores available for each participant: the CEFR level for listening in Swedish and the CEFR level for reading in Swedish.

2.2.3 Vocabulary test. To test the effect of the different types of subtitles on the

vocabulary acquisition of the participants, a vocabulary test was created. The participants made a pre- and a post-test, the vocabulary test was the same in the pre- and post-test. The test included thirty words in total. Twenty of these were occurred in the movie excerpt.

The test that Monique Bos (2014) designed that was based on the same movie except

could not be used without adjustments for the present experiment, as her participants were all

highly advanced learners and the audio language was not Swedish but English. Therefore, most

of the target words were too difficult for the beginners tested in the present study. Because there was also a pre-test, only using infrequent words in the test would be rather demotivating to the participants who would not know any of the words. Some of the original target words were lost in translation due to a free translation. Exact homographs in Dutch and Swedish were also excluded. One of these words was cutlass, which is sabel in both Dutch and Swedish.

However, ten of the words from the original test were suitable for the test. They are visible in the Table 3 below (nr. 1 - 10). The Swedish words, their translation to English and their translation in Dutch can be seen. The next ten words are words that also occurred in the movie excerpt and that were more suitable than the words in the original test (nr. 11 - 20). The final ten words (nr. 21 - 30) were taken from other parts of the movie that the participants in this study were not exposed to and were used as a control for the learning effect that occurs in pre-test post- test experiments.

Table 3 Overview of the vocabulary items that are used in the pre- and post-test: the Swedish word, the English word, the translation to Dutch and the difficulty of the word on the CEFR scale according to the Kelly-list. An asterix is used to mark words that did not occur on the Kelly-list.

Swedish English Dutch Difficulty (Kelly-list)

1 Attans Darned Donders, Verdorie *

2 Bläckfisk Squid Inktvis Bläck C2, fisk A2

3 Byte Booty Buit A2

4 Hissa Hoist Hijsen *

5 Hörru Avast/Hey Hé daar *

6 Ofattbara skatter

Unimaginable treasures/riches

Onvoorstelbare schatten

Ofattbar C1, skatt A1

7 Sånger Songs Liedjes A2

8 Sjögurkar Sea cucumber Zeekomkommers Sjö A2, gurka C2

Table 3 Continued

Swedish English Dutch Difficulty (Kelly-list)

9 Sjötulpan Sea tulip Zeepok/zeetulp Sjö A2, *

10 Trofé Trophy Trofee C2

11 Besättning Crew Bemanning B1

12 Dront Dodo Dodo *

13 Framgång Success Succes A1

14 Guld Gold Goud A2

15 Önskning Wish Wens Önskan B1

16 Segel Sail Zeil C1

17 Sjöröveri Piracy Zeeroverij *Sjö A2, förövare B2

18 Skepp Ship Schip A2

19 Skoj Fun Plezier/lol C1

20 Tillfälle Chance Kans A1

21 Besök Visit Bezoek A1

22 Glimmande ögon

Glistening eyes Glinsterende ogen * Öga A1

23 Matematik Mathematics Wiskunde B1

24 Meddelande Message Bericht A2

25 Pengar Money Geld A1

26 Plundrandet Plundering Plundering *

27 Seriemördare Serial killer Seriemoordenaar Serie A1, mördare B1 28 Skönhetssömn Beauty sleep Schoonheidsslaapje Skönhet B1, sömn B1

29 Snälla Please Alsjeblieft *

30 Stjäla Steal Stelen A2

Because the experiment was designed to compare the performance of the beginners and the more advanced learners, the vocabulary test had to be appropriate for people with a wide range of proficiency levels. The Swedish Kelly-list can be used to assess the difficulty of words for language learners. It is a vocabulary list that contains the 8425 lemmas that are used most frequently in modern, general Swedish. It based on the SweWAC corpus, that comprises 114 million words (Volodina & Johansson Kokkinakis, 2017).

For each word in the Kelly list, the CEFR level that corresponds to the vocabulary range that needs to be reached to know the word is stated. This level is also included in the table with the target words. As can be seen, the difficulty of the words ranges from A1 to C2. Some words were too specific to be included in the Kelly list, such as dront (“dodo”). These words will prevent the test from being too easy for advanced learners. This is necessary, for if they know all words in the pre-test, no effect of the subtitles can be measured. Four words were estimated to be within the vocabulary range of an A1 learner of Swedish, so even beginners would not have to despair. Because of the similarities between Swedish, Dutch and English, the estimation by the Kelly list is unlikely to totally correspond to the actual difficulty for the learners in this

experiment. For example, the word trofé is highly infrequent and labeled as C2, but the Dutch and English equivalents are trofee and trophy, which are quite similar in sound and orthography.

This could also be noticed by less advanced learners.

The participants’ word acquisition was tested using words that were extracted from the movie. This method of testing vocabulary was also applied in one of d’Ydewalles studies (d

’Ydewalle, 2002). The participants listened to the words and wrote down both the Swedish

spelling of the word and the meaning of the word (see appendix A). This way of testing was

preferred over exercises in which written target words in either Swedish or Dutch were used.

Using the Swedish spelling would give an advantage to the participants that watched the movie with intralingual subtitles. On the other hand, using Dutch would advantage the participants that watched the movie with interlingual subtitles and would not be possible to understand for the non-Dutch participants. The input that all participants had in common was the Swedish audio, so that was the most objective channel to use in the instruction. Only the passive acquisition was tested to make the test more feasible for beginners.

The target items were cut from the audio of the DVD. The resulting audio files were inserted into a Powerpoint presentation. Each file was uploaded on a different slide, so the Powerpoint presentation had thirty slides. To reduce the possible impact of the order of items, two different versions of the Powerpoint were made, in which the items were presented in a different order. Two different answer sheets were created, one for the pre-test and a slightly different one for the post-test, as the post-test included two content questions and a recognition exercise. There was room to write down the Swedish spelling of each word and the translation.

The post-test sheets also included two questions about the content of the movie, to test if the participants had understood the story. Besides, there was space to write down whether they recognized the word from the movie excerpt or not. Both of the sheets can be found in appendix A.

2.2.4 Movie excerpt. During the experiment, participants watched an 11 minutes and 24

seconds long movie excerpt with either Dutch or Swedish subtitles. The scenes were chapter

three and four of the movie ‘The Pirates! Band of Misfits’ (Lord & Newitt, 2012). This is an

animated movie produced by Aardman Animations and Sony Pictures Animation. This movie

was also used by Bos (2014) for her thesis that also investigated the effect of different types of

subtitles, which makes the studies more comparable. The reasons for choosing this movie was

that the movie is relatively unknown, but also funny and entertaining for the participants.

Furthermore, the occurrence of infrequent pirate vocabulary made the use suitable for vocabulary acquisition of the highly advanced participants in her study. Important for the present study was that there is also a Swedish version of the movie, in which Swedish voice actors were used. In Swedish the movie is called Piraterna! (“The pirates”) and that version was used in this study.

2.2.5 Subtitles. The subtitles that were used in the experiment had to be as accurate as possible. Because the original Swedish and Dutch subtitles were based on the English audio, there were many differences between the Swedish audio and the subtitles. Compare for instance the following two subtitle lines:

1.) Det är bara kommersiellt struntprat.

2.) Det är bara kommersiellt nonsens.

The first sentence is included in the subtitles that can be found on the DVD, and the second line corresponds with the Swedish audio and is used in the experiment. Although struntprat and nonsens are synonyms for “nonsense”, using the first line in the subtitles could be confusing for

the language learners. To avoid these and similar occurrences of synonyms as opposed to

identical words, the subtitles were adjusted to more accurately transcribe the Swedish voices. To write the subtitles, the computer program Subtitle Edit was used (Olsson, n.d.). While adjusting the subtitles, the following rules were taken into consideration, that were also used by Monique Bos and adapted from d’Ydewalle, van Rensbergen, and Pollet (1987) (Bos, 2014, p. 23):

 A subtitle can have a maximum of two lines

 A subtitle must include at least three characters

 A line of subtitles may not exceed 36 characters, including spaces.

 One-line subtitles should be displayed between 1.5 and 3.5 seconds.

 Two-line subtitles should be displayed between 1.5 and 6 seconds.

 There must be at least 0.25 seconds blank between different subtitles

The Swedish subtitles were made after carefully listening to the Swedish soundtrack, and were checked by a native speaker of Swedish. Sometimes the exact transcript had to be abridged due to the limited space and time. After the Swedish subtitles were completed, the Dutch

subtitles were written to accurately translate the audio. Both the Swedish subtitles for the intralingual condition and the Dutch subtitles for the interlingual condition can be found in appendix B.

2.2.6 Eye-tracker. Finally, an AVI file for each condition was uploaded on a computer that was attached to the Tobii T120 eye-tracker that was used during the experiment. The program Tobii Studio was used to record the eye movements of the participants (“Tobii Technology,”

2013). In that program, Areas of Interest (AOIs) in the movie were created, for which the eye tracking measures such as fixation time could be calculated.

Figure 9 Still of the movie excerpt. In this example the AOIs "Swedish subtitles" and "byte" (booty) are active and marked in red and yellow respectively

For each movie (the interlingual and the intralingual) the area of the screen that displayed

the subtitles was included. Furthermore, for each target word that appeared in the subtitles,