Learn Young, Learn Fair?
An investigation of the influence of Foreign Languages in the
Elementary School on the oral fluency of children
Stephana de Wolf
s1724525
MA thesis Applied Linguistics
Faculty of Arts
University of Groningen
22/08/2014
Word count: 16.708
Acknowledgements
There are a number of people without whom this thesis might not have been written, and to whom I am greatly indebted.
First of all, I would like to express my sincere gratitude to Bas Hemker, Jan van Weerden, Bertil Geurts, and Joke van Daal, employees of CITO, who were so generous to share their data with me. They provided audio fragments and data which I otherwise would not have been able to attend. Without their support, this thesis might not have been written. In addition, I would like to thank Elly Deelder and Eline Wassens for their help to find participating schools.
I also wish to express my deepest appreciation to my supervisor dr. Wander Lowie. I am grateful and indebted to him for his expert, sincere, and valuable guidance and encouragement extended to me. I would also like to thank my second reader dr. Kees de Bot for taking the time to read my thesis.
In addition, I would like to take this opportunity to record my sincere thanks to my parents, Bernard and Tineke, who have always believed in me and supported me throughout my entire studies. Their unceasing encouragement and support have shaped me into the woman I am today.
Last but not least, I would like to express my deepest thanks to the love of my life, Andrea, for the inspiration she has been to me. I would like to thank her for her support, her encouragement, her pep talks, and her faith in me. She has been of great value and without her statistical help and advice after proofreading this paper, I would never have been able to finish my thesis. Thank you for never stop believing in me and never letting me down.
List of Abbreviations
CAF Complexity, Accuracy, and Fluency
CEFR Common European Framework of Reference for Languages CLIL Content and Language Integrated Learning
CPH Critical Period Hypothesis
FLES Foreign Languages in the Elementary School
L1 First language
L2 Second language
L3 Third language
Table of Contents
Abstract ... 1
1. Introduction ... 2
2. Literature Background ... 4
2.1. History of English Education in the Netherlands ... 4
2.2. The earlier the better? ... 6
2.3. Foreign Languages in the Elementary School ... 9
2.4. Critical Evaluation of FLES ... 10
2.5. Complexity, Accuracy and Fluency ... 12
2.6. Why Fluency? ... 14 2.7. Purpose of study ... 16 3. Method ... 18 3.1. Introduction ... 18 3.2. Methodological Approach ... 18 3.3. Participants ... 19 3.4. Materials ... 19 3.5. Procedures ... 20
3.6. Design and Analyses ... 22
3.7. Results ... 24
3.7.1. Mann-Whitney U-test ... 24
3.7.2. Regression analysis for category attitude ... 24
3.7.3. Regression analysis for category home situation ... 27
3.7.4. Regression Analysis exposure outside school ... 29
4. Discussion ... 31
4.1. FLES and oral fluency ... 31
4.2. Influence of attitude and motivation towards English ... 32
4.3. Influence of the home situation ... 33
4.4. Influence of English exposure outside school ... 34
4.5. Compare and contrast ... 36
5. Conclusion ... 38
5.1. Limitations ... 39
6. Bibliography ... 40
1 ABSTRACT
Foreign languages in the elementary school (FLES) is currently receiving amount of attention in literature on second language learning and the Dutch government is also occupied with the topic. The present study has continued that focus by investigating whether oral fluency benefits from this early foreign language teaching. The participants were 29 non FLES students and 23 FLES students, all 10 to 12 years old and in the last grade of a Dutch elementary school (grade 8). They were tested on their oral fluency. The study used a picture description task of which the results of the audio fragments of the two groups were compared by means of a Mann-Whitney U-test. To investigate the influence of external factors like attitude and motivation towards English, language background, and exposure to English outside school, participants were asked to fill in a questionnaire. By means of regression analyses, this study also investigated the influence of external factors on oral fluency. The results were compared with CITO’s annual report, which stated that FLES students have better speaking skills than non FLES students. However, the present study did not find evidence to state that FLES students have a higher oral fluency than non FLES students. The results furthermore showed that generally speaking the external factors do not influence oral fluency either, although there exists some evidence that extracurricular activities like writing emails and listening to English songs contributes to the student’s oral fluency.
2 1. Introduction
On July 10, 2013 Sander Dekker, State Secretary of Education, announced that as of the year 2013-2014, Dutch primary schools were allowed to offer four hours of English language teaching from the first grade on (RTL Nieuws, 2013). Hitherto, the majority of Dutch primary schools generally taught English from the seventh grade on. This topic has gained great political attention recently, which makes it an interesting and highly relevant issue for research. Despite much diversity of opinion, Sander Dekker is positively disposed towards these Early Bird schools as a result of an experiment where children were taught other courses in English. According to Dekker, children have the capability to learn a second language with the greatest of ease, which should be turned to profit (RTL Nieuws, 2013). This experiment was requested by former Secretary of Education, Marja van Bijsterveldt (February, 23 2010 to November 5, 2012), who wanted universities to investigate the results of English in Dutch elementary schools. The results of these studies vary in outcome which leads to much diversity of opinions regarding Foreign Languages in the Elementary School (FLES). Some researchers argue it is best to start teaching a foreign language at a young age (before the age of 7) (Snow & Hoefnagel-Höhle, 1978; Clyne, Jenkins, et al 1995), whereas other researchers state that it is better to start teaching a foreign language when the child completely masters the mother tongue (which they believe to be at the age of 10 or 11) (Appel & Vermeer, 2001; Goorhuis-Brouwer & Schaerlaekens, 2000).
Since 2007, FLES has gained popularity in the Netherlands and the number of FLES schools started to increase. Ever since these schools started their projects, FLES has been exposed to a great deal of critical remarks. The main concern was whether FLES negatively influenced the children’s mother tongue (Goorhuis-Brouwer & De Bot, 2005). Researchers have been occupied with the topic and criticized the specific methods used for FLES as well as the role of the teacher. They also concluded that large numbers of external factors should be taken into account like the amount of English input as this highly influences the students’ output level. In addition, extramural exposure of English like music, travelling, and movies are assumed to highly influence students’ proficiency level, especially the student’s speaking skills. However, the influence of external factors on the student’s speaking skills have not been intensively scrutinized yet and is therefore highly interesting and relevant to investigate.
3 Hemker(2013) also showed that FLES students outperform non FLES students in vocabulary size. They furthermore investigated students’ speaking skills and concluded that FLES students have better speaking skills than non FLES students. However, their focus was primarily based on task-focused activities, known as ‘can-do’s’ which solely test the student’s ability to produce speech acts (e.g. greeting, thanking, warning, etcetera), whereas alternative present day evaluations of spoken production include complexity, accuracy, and fluency (CAF). Especially fluency is an important factor with regard to oral proficiency. Besides the CITO report by Geurts and Hemker (2013), the student’s speaking skills have not been scrutinized yet. Due to its additional complexity, oral proficiency is hard to investigate. Although it has been assumed that the students’ speaking skills will benefit most from FLES (Corda, Phielix, & Krijnen, 2012), it has not been convincingly demonstrated yet and leaves room for further research.
Due to the gap in literature, the primary purpose of the current study is to explore FLES and its influence on students’ oral fluency. The study also aims to investigate how external factors like the student’s attitude, home environment, and the amount of English exposure outside school contributes to children’s oral fluency. This paper furthermore attempts to contribute to an understanding of the influence of early language teaching on oral fluency. While a number of studies related to the present research have been undertaken since the rise of FLES, the focus of this earlier work was based predominantly on its effect on vocabulary size, writing, reading, and listening skills rather than oral fluency. This study is therefore occupied with the following research question: What is the effect of Foreign Languages in the Elementary School on the children’s oral fluency? To formulate a proper answer to the question, two groups of participants were involved: students of FLES schools and students of non FLES schools. They were asked to fill in a questionnaire and participate in a picture description task. Their results were compared by means of a Mann-Whitney U-test after which regression analyses were conducted to investigate the influence of external factors.
4 2. Literature Background
Due to the tremendous increase of globalization and expanding opportunities to communicate through new media, English has gained popularity and has become the most important language for communication. It is for this reason that English language teaching developed into an important topic of political debates. The Dutch government argues that children need to be capable to communicate in English. Although the Dutch government is enthusiastic about the opportunity to offer English from the first grade on, the effects of Foreign Languages at the Elementary School (FLES) are not clear yet. It is interesting that despite the government’s ambitions, the effects of FLES on speaking skills have not been intensively investigated yet. The focus of this study is therefore on the effect of FLES on speaking skills, emphasizing the student’s oral fluency. In order to highlight the political essence of this study, understanding of the historical context of English in Dutch primary education is necessary. Subsequently, FLES will be critically evaluated. The then following sections will elaborate on the aspects of fluency.
2.1. History of English Education in the Netherlands
The debate about second language teaching in Dutch primary education arose in 1972 (Corda, Philipsen, & de Graaff, 2014). The majority of the Dutch government preferred English teaching as a second language, whereas the minority preferred French or German over English. People came to realize that English gained a stronger position as a language for international communication. In 1980, the Dutch Parliament based their choice for English on the arguments that English functioned as a lingua franca in many parts of the world. In addition, English is the first foreign language in Western-European countries. The members of the Dutch parliament also stated that teaching English as a foreign language at primary schools contributes to a solid basis of English on which pupils can build at secondary school. As a last argument the Dutch parliament emphasized the fact that over 90% of the students in secondary education choose English as a final examination subject. This indicates that English dovetails with general and practical needs (Corda, Philipsen, & de Graaff, 2014).
5 situations. They were furthermore required to develop a positive attitude towards learning foreign languages rather than gathering knowledge about the language. The Dutch government decided that 80-100 hours in grade 7 and 8 would be sufficient to achieve this goal (Van Toorenburg & Bodde-Alderlieste, 2003). The government however, has never given further foundations for the required 80-100 hours which makes it seem like it was an arbitrary choice and therefore an ill-considered decision.
It was in 1993 when these guidelines became official and it then became obligatory for schools to offer English language teaching (Gidding, 2011). However, schools were free to decide what method they wanted to use for teaching English as a foreign language. Additionally, schools were free to make their own choices with regard to the didactic approach, what grade they would start, the teacher’s requirements, and the amount of time they wanted to spend on English language teaching (Geurts & Hemker, 2013). Despite the guidelines, schools were exposed to plenty of freedom and as a result of improper functioning of the schools inspectorate, it remained unclear whether the main goals were attained. A study by CITO investigated the current situation of English teaching at Dutch elementary schools (Geurts & Hemker, 2013). The report showed that until 2006, Dutch primary schools came short of the 80-100 hours English language teaching a week. Only 45 minutes per week was spent on English teaching, meaning that children were only provided with 60 hours of English lessons during the last two grades of primary school (Geurts & Hemker, 2013). Despite the official guidelines, the guidelines remained directions rather than official requirements. Due to the lack of official requirements, no sanctions could be imposed on schools that came short of the 80-100 hours, which can be considered as a serious shortcoming.
6 Wetenschap, 2004: 7). In order to achieve this, the then secretary of Education stated it is best to start teaching a foreign language at a young age, meaning before the age of 10-11.
2.2. The earlier the better?
People tend to believe that the younger you start learning a second language, the better. This idea derives from the observation that abilities to acquire new skills diminish with age (Hakuta, Bialystok, & Wiley, 2003). When assuming this observation holds good, it could be stated that such an age effect also exists in second language learning. In other words: the older you get, the harder it becomes to learn a second language. Hakuta et al. (2003) believe that there is an age-related decline in the success with which individuals master a second language. However, there has been an ongoing debate for ages on the topic of second language learning. Although ‘learn young, learn fair’ has gained universal scientific acceptance, researchers have not found mutual consent on the perfect age to start learning a second language yet.
Clyne, Jenkins, Chen, Tsokalidou, and Wallner (1995) present several reasons to start teaching English at an early age. They state that early language teaching has advantages as children who have become familiar with a second language at a young age (between 0 and 7 years old) develop their metalinguistic consciousness which means they are better at reflecting on and thinking about the characteristics and functions of a language. Additionally, Clyne et al. (1995) argue that children starting to learn a second language at a very young age, have a better pronunciation in the target language as young children are good at imitating sounds. Moreover, young children show a higher psychological flexibility than adults, meaning they are better in putting themselves in someone else’s position. As a last advantage Clyne et al (1995) mention that the earlier children start acquiring a second language, the more time they have to learn that particular language. In addition, Snow and Hoefnagel-Höhle (1978) argue that second language acquisition will be relatively fast and successful if it occurs before the age of puberty.
In contrast to Clyne et al. (1995), Nikolov and Djigunovic (2006) state that adolescents are perfectly capable of learning a second language. In their studies, Appel and Vermeer (2001) and Goorhuis-Brouwer and Schaerlaekens (2000) also state that adults do better than young children in the early stages of second language acquisition. Researchers believe this is due to the prior knowledge they have obtained in their first language (L1): older children understand the system of the L2 quicker and better and are able to consciously learn new words and rules (Appel & Vermeer, 2001).
7 research in both first and second language acquisition. Rather than elaborating on the observation that abilities to learn a new language diminishes with age, the CPH refers to a particular period during which a second language is acquired best. According to this hypothesis, a biologically-based critical period for first and second language learners exists that prevents older learners from achieving native-like competences (Hakuta, Bialystok, & Wiley, 2003). It is argued that a critical period consists of two characteristics, namely “a high level of preparedness for learning within a specified developmental period to assure the domain is mastered by the species, and a lack of preparedness outside of this period” (Hakuta, Bialystok, & Wiley, 2003, p. 3). Strictly speaking, outside of the critical period a different relationship exists between age and learning than inside of the critical period.
The CPH was firstly proposed by Lenneberg (1967) who argued that primary language acquisition must occur during a critical period which ends approximately at the age of puberty with the establishment of cerebral lateralization of function. Any language acquisition which takes place after the age of puberty will be different or even slower and less successful than normal first language learning (Lenneberg, 1967). It is thus assumed that children are better second language learners than adults because their brains are specially organized to learn a language, whereas those of adults are not (Birdsong, 1992). More evidence in favor of this critical period hypothesis comes from several sources. In addition to Lenneberg, Seliger (1978) and Walsh and Diller (1981) argued that there are more critical periods and not just one critical period affecting all aspects of language at the same time. Many critical periods close off different abilities. The first ability that easily gets lost is the one to master a native accent in a foreign language, which occurs around the onset of puberty. Furthermore, Scovel (1988) believes that a critical period only applies to the acquisition of pronunciation of a second language. He additionally states that second language learners who start later than the age of 12, will never be able to reach a native like accent (Scovel, 1988).
In this manner, the CPH is an idea of a critical period after which it is no longer possible for L2 learners to reach a native-like level, especially for pronunciation. However, some researchers have argued that the evidence fails to support this interpretation. As a first argument researchers emphasize the identification of older learners who achieve native-like competences in the target language (Birdsong 1992; Bongaerts, Planken, & Schils, 1995). They furthermore concluded that individuals may attain a native like accent, in spite of a late start (Birdsong 1992; Bongaerts, et al., 1995). The second argument Hakuta et al. (2003) give, is the behavioral evidence that was unsuccessful to show a qualitative change in learning outcomes at the close of a critical period.
8 age of 50 than at the age of 14) it is questionable whether this points to an age effect or to a critical period. Bialystok and Hakuta (1999) believe that there is not enough evidence to state that there exists such a thing as the critical period hypothesis. However, they did find that the age at which people start learning a second language corresponds in some way to the ultimate success that the person will attain after years of having used that language. Crucial point in the CPH debate is that the critical period calls for specifying the point in time. However, researchers have not reached consensus about what age constitutes this critical point in time (Hakuta et al., 2003). Claims about the age at which the critical period ends, have varied from five years (Krashen, 1973), six years (Pinker, 1994), 12 years (Lenneberg, 1967; Scovel, 1988), and 15 years (Johnson & Newport, 1989). Hakuta et al. (2003) used both 15 and 20 years as a cut-off point for the end of the critical period, yet no evidence of a change in language learning potential at those times, was found. As long as no consensus is reached about the cutoff point, the construct of a critical period is not convincing.
Additionally, Snow and Hoefnagel-Höhle (1978) state that second language learning up to the age of 5 may be slower than after the age of 5, due to the fact that the effort needed to learn a first language is disadvantageous to second language acquisition. Furthermore, Ervin-Tripp (1974) stated that having already learned a language makes it easier to learn a second language. In other words: the child first needs to fully control the first language. Although late learners are sometimes able to achieve native-like perfection in a second language and experimental results sometimes show late learners performing just as well as early learners, the older group on average performs worse (Birdsong, 1992). Despite the fact that the CPH has been the praised theory with regard to the topic of second language learning, the contradictory findings do not make it a reliable theory.
9 found evidence that exposure to English outside the classroom influences second language acquisition as well.
2.3. Foreign Languages in the Elementary School
Although no consensus has been reached among researchers with regard to the existence of a critical period, basically all researchers agree that learning most skills is easier at younger age, especially complex skills. Although this is not identical to a critical period, the age effect is still there. The Dutch government therefore holds on to the idea that the earlier you start learning a language, the better. As a result of the European ambitions and Dutch policy based on the ambition that citizens should be fluent in two other languages besides the mother tongue, ever more schools offering English from the first grade on, started to develop. In June 2003, a school in Rotterdam took the initiative to start with ‘Early English’. The project named Early Bird was considered successful and more schools followed. Early Bird has gained popularity and success outside Rotterdam, resulting in a network with approximately 250 associated schools. In 2008, the Dutch Advisory Council for Education underlined the importance of the European ambitions that (Dutch) children should achieve an appropriate level in two foreign languages. From this moment on, FLES started to grow. In order to achieve an appropriate level of English, researchers stated more time should be spent on English language teaching than the set 45 minutes per week. As a consequence, in 2009 the Dutch Department of Education agreed to start an experiment which implied that 15% of the lessons (four hours a week) had to be taught in English under the auspices of the European Platform (Van Loon & Setz, 2012). Thirteen schools participated in this project. The project has awakened enthusiasm which is reflected by the increasing number of FLES-schools: whereas there were only 15 English FLES schools in 2003, there were 983 in 2013 (see appendix A) (Corda, Philipsen, & de Graaff, 2014). Although the number of FLES schools has increased, the pilot project has also demonstrated that schools do not meet the required four hours of English language teaching a week. Goorhuis-Brouwer and Schaerlaekens (2000) state that children in first and second grade need to be provided with a high amount of authentic input. If the required four hours is diminished, it is questionable whether FLES will indeed have the desired effect (Corda,et al., 2012; Thijs, Trimbos, Tuin, Bodde, & De Graaff, 2011; Van Loon & Setz, 2012).
10 to the fourth grade (Thijs, Tuin, & Trimbos, 2011). Sander Dekker attempted to stimulate schools to start teaching English as early as possible and offer more hours of English teaching a week in order to achieve a high level of English proficiency. At the majority of non FLES schools, English language teaching is restricted to 45 minutes per week. According to Dekker, these 45 minutes should be increased and he therefore recommended teaching other courses like history and geography in English in order to achieve a higher amount of English input. This integrated way of language teaching is also known as CLIL: Content and Language Integrated Learning.
2.4. Critical Evaluation of FLES
Despite its breakthrough, FLES has been a controversial topic leading to emotional debates. Although it has become the ultimate way where both English is taught and (some) other subjects are taught in English at the beginning of elementary school from first grade on, there still are a lot of catches. Many researchers have investigated the effects of FLES and the results are diverse. One of the reasons leading to differing research outcomes is that only a limited number of schools in the Netherlands start teaching English in the first grade (at the age of four), which makes it complex to take stock of FLES’ reliability. Additionally, as these schools have only started recently with FLES, there are no subjects who have finished the entire eight-year course, which hampers reliable research. Studies can only investigate the effects of FLES at the end of the first grade, which may not seem to be representative. Also, the schools that have started teaching English in the third grade which have finished the course, is only a limited number.
11 progress. This automatically leads to another issue that needs to be addressed, namely from how many hours per week FLES starts being useful. There seems to be agreement on the idea that second language teaching at a young age will only yield fruit when the amount of (authentic) input is high. Young children need more time than older children to achieve the same proficiency level (Burstall et al., 1974; Krashen et al., 1979). Goorhuis-Brouwer and Schaerlaekens (2000) state that children in first and second grade need to be provided with a high amount of authentic input. Moreover, they believe that grade three to six are less suitable for English language teaching, as they simultaneously develop Dutch writing skills and mathematic skills. These statements are not in line with what the government believes.
Additionally, researchers have not reached consensus on the conditions of FLES education. There has not been found a didactic approach that works best. Herder and De Bot (2007) believe that different age groups require different teaching methods. In their article, they refer to Gilzow and Rhodes (2000) and Codina and Smits (2001) who distinguished several didactic approaches. With the total immersion approach, 50-100% of the instruction period is taught in the target language. The 2010 pilot project has already demonstrated that teaching 15% of the teaching period in English is hardly attainable, so this didactic approach does not seem appropriate for Dutch elementary schools. The same holds for the partial immersion approach and the two-way immersion approach, where 50% of the instruction time is taught in the target language. Extra hours of English teaching are often filled in with grammar instruction, which might not be a successful approach. Researchers tend to agree that second language learners need to be provided with authentic material and CLIL is suggested as a suitable method. In 2010, a pilot project was launched which gained permission to teach 15% of the education time in English, which equals four hours a week. However, participating schools were not able to meet the required 15%. This leads to the question addressed before, whether this limited amount of English input actively contributes to or improves the pupil’s level of English.
12 into the pupil’s daily routine. As children in the first two years of elementary school receive English in a receptive and communicative way (Herder & de Bot, 2007), Goorhuis-Brouwer (2007) advocates two teachers for FLES: A non-native speaker who speaks Dutch with the children and a native-speaker who speaks English with them. It has also been suggested by Groot (2006) to make use of a language assistant. The results of the pilot project (Van Loon & Setz, 2012) showed that the help of language specialists is extremely meaningful.
Although researchers believe that the help of language specialists is extremely helpful, the few studies have not found a proper solution to the ‘FLES teacher problem’ yet. Even though Meijerink (2012) emphasized that a decent level of English proficiency is required, the annual report by CITO (Geurts & Hemker, 2013) demonstrated that the teacher’s English level leaves much to be desired. In 2012, 85% of the teachers at Dutch elementary schools also provides English lessons. These teachers are variously trained and have achieved English in sundry manners. A worrisome aspect is that only a small number of the teachers at Dutch elementary schools (11% at non FLES schools and 0% at FLES schools) are satisfied with the extent of English at the Teachers’ Training College for Primary Education (Dutch PABO). It is even more alarming that 59% of the teachers did not have English lessons during their studies at the Teachers’ Training College for Primary Education (Geurts & Hemker, 2013).
2.5. Complexity, Accuracy and Fluency
Due to varying opinions on FLES and diverging results published recently, this present study investigates the success of FLES again. Whereas the influence of FLES on vocabulary and proficiency has already been investigated (Naber, 2011; Unsworth, De Bot, Persson, & Prins, 2012), the influence of FLES on speaking skills has not been extensively investigated yet. The annual report by CITO (Geurts & Hemker, 2013) did investigate the student’s speaking skills, yet their study solely focused on task-focused activities, known as ‘can-do’s’. Can-do’s tend to test the working memory rather than the capability to speak, as it solely focuses on the ability to produce speech tasks. However, alternative present day evaluations of spoken production include complexity, accuracy, and fluency (CAF).
13 been calls for means by which to gauge second language development. The first call came from Hakuta (1976). Hakuta was interested in tracking his subject Uguisu’s English language development over time, as Brown (1973) and other L1 acquisition researchers had done for children learning English as a native language (Larsen-Freeman, 2009). In Brown’s study, researchers relied on the mean length of utterance as a measure of development. These calls indicated the absence of a suitable developmental yardstick for the construction of an SLA index of development for (spoken) production.
Traditionally, researchers (and people in general, also in school contexts) tend to focus on the learner’s L2 accuracy. An attempt to break away from this perspective was to objectively evaluate different dimensions of proficiency, yielding a more complete picture than solely focusing on errors. This attempt was found in the CAF-approach. CAF has figured as major research variables in applied linguistic research. During the 1980s, the focus of many linguistic research was on L2 pedagogy (Housen & Kuiken, 2009). Soon a distinction was made between fluent versus accurate L2 usage to investigate the development of oral L2 proficiency in classroom contexts. Brumfit (1984) was one of the first to use this division. He distinguished between fluency-oriented activities, and accuracy-oriented activities. The first focuses on spontaneous oral L2 production, whereas the linguistic form and the controlled production of grammatically correct linguistic structures in the L2 is the focus of the latter. In the 1990s, a third component was added to the triad. By adding complexity, Skehan (1989) was followed who proposed an L2 model which for the first time included CAF as the three principal proficiency dimensions. In the 1990s, the three dimensions were also given their traditional working definitions, which are still used today.
According to Housen and Kuiken (2009), accuracy is the oldest dimension of the triad. They argue that it also is the most transparent and consistent construct of the triad. Wolfe-Quintero et al. (1998) refer to accuracy as the ability of the L2 learner to produce error free speech and the “conformity of second language knowledge to target language norms” (Wolfe-Quintero et al. 1998, p. 4). Additionally, accuracy means referring to the degree of deviancy form a particular norm and deviations from the norm are commonly referred to as errors.
14 Linguistic complexity has been commonly interpreted as the size, elaborateness, richness, and diversity of the learner’s linguistic L2 system while cognitive complexity is seen as a broader notion.
Fluency, on the other hand, is characterized by “perceptions of ease, eloquence, and ‘smoothness’ of speech or writing” (Housen & Kuiken, 2009, p. 463) and therefore refers to a person’s general language proficiency. Language researchers have analyzed oral production data to determine exactly which quantifiable linguistic phenomena contribute to perceptions of fluency in L2 speech. The broad finding was that speech fluency is a multi-componential construct in which different elements can be distinguished, such as breakdown fluency (number, length, and distribution of pauses in speech), speed fluency (rate and density of delivery), and repair fluency (number of false starts and repetitions). Fluency is thus related to learners’ control over their linguistic L2 knowledge, as reflected in the speed and ease with which they access relevant L2 information to communicate meanings in real time, with control improving as the learner automatizes the process of gaining access.
2.6. Why Fluency?
The present study focuses on concept fluency and the reason for that is evident: the English level of the participants is remarkably low which makes it impossible to focus on accuracy or complexity. As mentioned before, accuracy is described as the ability of the L2 learner to produce error free speech. In many cases, their speech is a string of errors. For this reason, accuracy is left aside. In addition, complexity focuses on relative difficulty with which language features are processed in L2 performance. The participants in this study are 10 to 12-year-olds whose English is extremely basic. There is no use in testing their accuracy nor complexity. Furthermore, the aspects of spoken production have been tested by CITO. Fluency on the other hand has not been tested before and is therefore the main focus of this study.
15 many people would like to perform better in their L2. In order to understand this question, Segalowitz (2010) points out why it is so difficult for people to become fluent in another language if they are already fluent in their L1. This question is referred to within-individual fluency (Segalowitz, 2010, p. 2). Another issue Segalowitz (2010) addresses, is the need to understand why between-individual differences exist. Segalowitz tries to find an explanation why some between-individuals perform better in an L2 than others. According to Segalowitz, answering these questions will lead to identifying practical steps to reduce unwanted fluency gaps.
In order to understand these fluency gaps and answer the related questions, it first needs to be clear what fluency actually means. Whereas the term oral fluency has been widely used with respect to second language speech, no generally agreed definition of the term has been established yet. Oral fluency is a concept used to indicate different constructs. In everyday language-use oral fluency refers to overall language performance in which native-like performance is considered the ultimate goal (Cucchiarini, et al., 2002). Many researchers have been occupied with defining oral fluency. Fillmore (1979) performed one of the first studies investigating fluency. He first defined fluency as the ability to talk at length with few pauses and to the ability to fill the time with talk. In addition, the speakers should be able to deliver the message in a coherent way and without hesitations. He then stated that a person can only be considered fluent if the speaker knows what to say in a wide range of contexts. Fillmore (1979) finally argued that fluent speakers use creative and imaginative language. As a maximally fluent speaker, Fillmore (1979) refers to a speaker that meets all above mentioned requirements.
16 In a narrower sense, fluency is formulated as just one component of oral proficiency (Cucchiarini et al., 2002). Segalowitz (2010) refers to three aspects of fluency, namely cognitive fluency, perceived fluency and utterance fluency. Cognitive fluency can be defined as the ability of the L2 speaker to smoothly translate thoughts to L2 speech. Perceived fluency refers to the subjective measure of what listeners perceive, which is about the L2 speaker’s cognitive fluency. Utterance fluency is summarized as objective acoustic measure of an utterance.
It is not solely the definition of fluency that has been topic of debates, but its measurement as well. Judging the candidates’ fluency is a common approach in language testing. However, more reliable criteria needed to be developed now that it has been discovered what temporal measures of fluency contribute best to the listeners’ perception of fluency and distinguish fluency an non-fluent speakers. Cucchiarini et al. (2002) state that many studies that have been conducted to investigate which variables contribute to perceptions of fluency, suffer from methodological shortcomings. The main reason for failing is due to the very few subjects in the studies and the fail to employ reliable methods of analyzing the duration of pauses.
2.7. Purpose of study
Briefly summarized, it can be stated that since the increasing popularity of FLES, many researchers have been occupied with the topic of early language teaching. The Dutch government has set goals based on European ambitions that all European citizens should at least acquire two other languages besides the mother tongue. The European Council has based these ambitions on theories that state it is best to start learning a foreign language as early as possible, as there is a certain cut-off point after which the process of second language learning becomes less slowly. However, the underlying theory (CPH) is unsettled as the researchers cannot reach consensus. Moreover, the findings of the effect of FLES vary. The Dutch government has made decisions while the topic is still open to question. Many questions regarding the topic of FLES are still unclear and there is a gulf between theory and practice.
17 What is the effect of Foreign Languages in the Elementary School on children’s oral fluency?
The reason for solely focusing on speaking skills, narrowing down to fluency, is that current research has not elaborated on the effect of FLES on oral fluency yet. Although it has been assumed that FLES may have a positive effect on the children’s speaking skills, there is no convincing evidence in literature for this assumption. The main reason for this lack in literature is that oral fluency is hard and complex to investigate. First of all, fluency is a broad and complex concept, consisting of various variables. Furthermore, factors such as the student’s language background, the student’s attitude towards the English language, and the exposure to English outside school need to be taken into account are. This leads to the following sub questions:
(1) To what extent is the student’s attitude and motivation towards English of influence of the oral fluency?
(2) To what extent does the student’s home environment influence the student’s oral fluency?
(3) To what extent does exposure to English outside school contribute to the oral fluency?
The answers to the sub question will hopefully contribute to giving a solid, complete, and appropriate answer to the main question.
18 3. Method
3.1. Introduction
This chapter introduces and discusses the methodological approach and research design best suited to examine the research question set out in chapter 1. In order to answer the question What the effect of Foreign Languages in the Elementary School is on the child’s oral fluency, 52 audio fragments of 52 Dutch students have been analyzed. These audio fragments were provided by CITO. For their annual report in 2013 they needed many participants of both FLES schools and non FLES schools, to draw up the balance sheet for the then state of affairs (Geurts & Hemker, 2013). Under strict terms and conditions, CITO shared their data for the purpose of this study.
In this chapter, a qualitative research method is proposed to arrive at answers to the research question. An overview of the research design then follows, beginning with an outline of the key methods employed. Given the importance of design and validity in the choice of research instruments, justification of each method used is provided. The subsequent section includes an illustration of the specific process of data collection, followed by an overview of methods used for data analysis. The results of this data analysis will be discussed in the next section after which the results will be interpreted. The chapter concludes with a brief summary of the preceding sections.
3.2. Methodological Approach
19 proposed measures by De Jong (2013). Breakdown fluency is related to hesitation phenomena (Segalowitz, 2010) and De Jong (2013) advocates speech rate, mean length of utterance in syllables, mean length of utterance in seconds, number of pauses per minute in total time, number of pauses per minute in speaking time, mean pause duration, and phonation time ratio as appropriate measures.
3.3. Participants
In order to investigate the differences between FLES students’ and non FLES students’ oral fluency, two distinct groups were selected. These groups were selected and tested by CITO in 2012 and were drawn from different elementary schools in the Netherlands. Five FLES schools and six non FLES schools participated in this study and all schools sent a few students to take part in the activity. The participants were divided into a FLES group and a non FLES group. All participants were in the eighth grade and were therefore between ten and twelve years old (Geurts & Hemker, 2013).
The non FLES group existed of 29 participants and were students of different elementary schools in the Netherlands. Participating schools were randomly selected by CITO. A crucial criterion for division into groups during the selection procedure was the starting year of English language teaching. Schools in this group started teaching English in grade 7. Out of 29 participants of the non-FLES group, 14 participants were male and 15 participants were female. These participants had only had English as a subject in elementary school for two years. 24 out of 29 participants have Dutch as their L1, the remaining 5 participants either have a mix or a foreign language as their L1, varying between Chinese, German and Frisian.
The second group existed of 23 FLES students, meaning these participants had had English from the first grade on. The participating schools were chosen by CITO and selected based on the starting year of English language teaching. The group existed of 8 male and 15 female participants. 20 participants have Dutch as their L1 and 3 of the participants either have a mix or a foreign language as their L1, varying between Moroccan and English.
3.4. Materials
20 constructions in their mouths, and without making demands on memory for story events encountered before. In this way, the participant delivers natural speech which is the main motivation for this choice of speech elicitation tasks. Additionally, Kormos and Dénes (2004) argue that interactive tasks may complicate computer-analysis of speech phenomena as it might happen that the two speakers talk simultaneously. Kormos and Dénes (2004) also state that picture description task is favorable as the fluency of the production can be influenced if the speaker has to produce different types of content. Different types of content places different cognitive load on speakers. With the use of a picture description task a fixed content is provided, which could eliminate the influencing factor of content. The cartoon used in this study was chosen by CITO and consisted of six pictures arranged in a logical order and can be found in the appendix. Selection criteria for this cartoon included relative simplicity of the story and of the vocabulary necessary to describe the story.
The audio fragments produced several speech samples which were analyzed on the following variables: speech rate, articulation rate, average syllable duration, mean length of utterance in syllables, mean length of utterance in seconds, number of pauses per minute in total time, number of pauses per minute in speaking time, mean pause duration, phonation time ratio, number of filled pauses and the dysfluencies per minute.
3.5. Procedures
For the picture description task, the time devoted to planning was specified to five minutes. These five minutes were provided for the participants to have enough time to understand the story depicted in the cartoon and to gather their thoughts about how they were to narrate it. The cartoon was shown on a laptop and participants had to turn pages manually. As different schools were participating, circumstances were not identical. In some occasions the experiment was taken in an isolated area whereas other recordings were taken in the classroom. The instructor was always present at the recording and provided the first sentence the participants had to start with. The students’ performance was recorded by laptop and were labeled with a school number and a student number, so the information could be analyzed anonymously.
21 In order to gain precise temporal measures, the speech samples were analyzed with the help of PRAAT computer software (Boersma & Weenink, 2013). PRAAT is a software program that with the help of a script successfully detects the total duration of the sample, the number of syllables, the number of pauses, speech rate, phonation time, average syllable duration, and the articulation rate. In 2009, De Jong and Wempe presented a script written in the software program PRAAT that automatically detects syllable nuclei in order to calculate speech rate. They used the nucleus of a syllable as the central part of a syllable (De Jong & Wempe, 2009). The script takes sound files as input and writes a TextGrid file with syllable nuclei marked in a point tier. In their article, De Jong and Wempe (2009) found correlations between human-measured speech rate and automatically measured speech rate. However, the script misses mostly unstressed syllables that are detected by human judges. Despite these missing syllables, correlations suggested that the algorithm works well in predicting the actual number of syllables. As De Jong and Wempe (2009) concluded that this script suffices for the purpose of measuring speech rate as number of syllables per time unit when comparing speakers and tasks, it is therefore used as a reliable measurement in this study.
For the audio fragments used in this study, the automatic settings of the script written by De Jong and Wempe (2009) in PRAAT were used: the silence threshold was set at -25 dB and the minimum pause duration was set at 0.3 seconds. The minimum dip between peaks deviated from the original settings. In order to avoid that PRAAT were to add non existing syllables, the minimum dip between peaks was set at 4 dB rather than the original 2 dB.
22 3.6. Design and Analyses
For this study, several steps were taken to arrange and analyze the data. First of all, the speech samples provided by CITO needed to be transformed into measurable data. PRAAT was used to transform speech samples into data.
The second step was the examining of the dependent variables that determine utterance fluency. In order to measure utterance fluency, the following temporal variables were examined:
1. Speech rate
Speech rate was calculated automatically by the PRAAT computer software, simply by importing the audio file to the program. PRAAT automatically detects the number of syllables. The number of syllables produced in a given speech sample was divided by the total duration of the speech sample (including pause time).
2. Articulation rate
In order to calculate the articulation rate, the number of syllables was divided by the phonation time. PRAAT automatically calculated this.
3. Mean syllable duration
The mean syllable duration (average syllable duration) was calculated by dividing the phonation time by the number of syllables. PRAAT automatically calculated the mean syllable duration.
4. Mean length of utterance in syllables
The mean length of utterance in syllables had to be calculated manually. The number of syllables was divided by the number of pauses plus 1.
5. Mean length of utterance in seconds
The mean length of utterance in seconds also had to be calculated manually. The speaking time was divided by the number of pauses plus 1.
6. Number of pauses per minute (total time)
The number of pauses were given by PRAAT. In order to calculate the number of pauses per minute in total time, the number of pauses were divided by the total time of the speech sample.
7. Number of pauses per minute (speaking time)
The number of pauses were also needed to calculate the number of pauses per minute speaking time. Here, the number of syllables were divided by the speaking time (phonation time).
8. Phonation-rime Ratio
23 phonation time was calculated by PRAAT automatically. The total number of syllables produced in the speech sample was divided by the amount of time taken to produce them in seconds. Pause time however, is excluded. All semantic units were counted, including filled pauses, hesitations, repetitions and partial words. The phonation-time ratio was calculated as the percentage of time spent king of the total duration of the speech sample.
9. Filled pauses per minute (speaking time)
The filled pauses per minute speaking time had to be counted manually. All filled pauses such as uhm, er, mm were divided by the speaking time.
10. Dysfluencies per minute (speaking time)
The total number of dysfluencies such as repetitions, restarts and repairs were divided by speaking time. This had to be calculated manually.
11. Mean pause duration
In order to calculate the mean pause duration, the total time minus the phonation time was divided by the number of silent pauses.
After all these measures had been calculated, FLES and non FLES students were compared on all eleven fluency variables by means of the Mann-Whitney U-test. The Mann-Whitney U-test is conducted as it is not assumed that the dependent variables are normally distributed. The Shapiro Wilk test showed that none of the variables are normally distributed. The null hypothesis for this statistical analysis is as follows:
H0 = non FLES = FLES
In other words, when the null hypothesis is accepted, it can be argued that non FLES students equal FLES students in their performances.
However, besides the differences between FLES and non FLES groups, other external factors may influence oral fluency as well. In order to control other variables, the questionnaire filled in by the participants, was analyzed. In this study, the external factors were categorized into three groups: attitude towards English, home situation, and the exposure of English outside school. An overview of these independent variables can be found in the appendix D.
24 (1) Oral Fluency = c + β1FLES + β2L1DUTCH + β3L1MIX + β4L1FOR + β5HLDUTCH + β6HLMIX +
β7HLFOR + β8SHLDUTCH + β9SHLENG + β10SHLOTHER + β11SHLDUTCHENG +
β12SHLDUTCHOTHER
(2) Oral Fluency = c + β1FLES + β2FUN + β3IMPORTANT + β4DVD + β5DARE + β6BOOKS +
β7INTERNET
(3) Oral Fluency = c + β1FLES + β2MAG + β3EMAIL + β4LET + β5HOL + β6SPEAKING + β7VIDEO +
β8GAMES + β9BOOK + β10MOVSUB + β11SONGS + β12INFO + β13TVSUB
An overview of abbreviations can be found in appendix E.
3.7. Results
3.7.1. Mann-Whitney U-test
The groups FLES and non FLES were compared by means of the Mann-Whitney U-test. The group with the higher rank can be determined by looking at how the actual rank sums compare to the expected rank sums under the null hypothesis. As can be seen in table 1, for two variables (mean length of utterance in seconds and filled pauses per minute speaking time) the P-value is smaller than α=0.05 and thus significant. In these two cases, H0 will be rejected. For the variables mean length of
utterance in syllables, number of pauses per minute speaking time speech rate, articulation rate, mean syllable duration, number of pauses per minute total time, mean pause duration, phonation time ratio, and dysfluencies per minute, H0 will be accepted, which means that non FLES equals FLES.
3.7.2. Regression analysis for category attitude
Table 2 shows the regression analysis for the category attitude and motivation towards English. The table demonstrates that the more positive the participant’s attitude towards reading English books and English messages on the internet is, the higher the speech rate is (regression 1). Regression 1-4 and 1-5 additionally illustrate that the more positive the participant’s attitude is towards reading an English book, the higher the mean length of utterance is in both syllables and seconds.
25 significant for the variable phonation time ratio, as shown by regression 1-8. The more positive this attitude, the higher the phonation time ratio. The last regression shows that a positive attitude towards English messages on the internet results in a lower mean pause duration.
Table 1. Comparison of the linguistic measures of non FLES and FLES students
Variable Group Obs. Mean Sd. Min Max Z p
Speech Rate Non FLES
FLES 29 23 1.539 1.706 0.669 0.700 0.360 0.320 3.040 3.080 0.866 0.3865
Articulation Rate Non FLES FLES 29 23 3.338 3.395 0.337 0.440 2.740 2.810 3.930 4.650 0.304 0.7611
Mean Syllable Duration Non FLES FLES 29 23 0.302 0.299 0.030 0.035 0.255 0.215 0.365 0.356 -0.341 0.7332 Mean Length of Utterance in syllable Non FLES FLES 29 23 4.404 5.493 2.114 3.178 2.200 3.038 9.429 16.500 1.806 0.0710 Mean Length of Utterance in seconds Non FLES FLES 29 23 1.298 1.602 0.542 0.875 0.672 0.876 2.736 4.973 1.981 0.0476**
Number of pauses per minute (total time)
Non FLES FLES 29 23 0.341 0.321 0.094 0.119 0.132 0.940 0.526 0.511 -0.341 0.7332
Number of pauses per minute (speaking time)
Non FLES FLES 29 23 0.830 0.692 0.266 0.231 0.320 0.168 1.339 1.074 -1.870 0.0615
Mean Pause Duration Non FLES FLES 29 23 1.858 2.069 1.229 1.930 0.720 0.627 6.570 9.621 -0.304 0.7611
Phonation time ratio Non FLES FLES 29 23 0.451 0.498 0.163 0.182 0.130 0.100 0.785 0.750 0.709 0.4781
Filled pauses per minute (speaking time) Non FLES FLES 29 23 0.220 0.104 0.209 0.090 0 0 0.893 0.314 -2.071 0.0384**
Dysfluencies per minute (speaking time) Non FLES FLES 29 23 0.096 0.097 0.067 0.074 0 0 0.220 0.277 -0.157 0.8754
Asterisks indicate statistical significance at the ***1 and **5 percent level.
26 Table 2. Regression analysis for the control variable attitude
Number in parentheses are absolute values of t-statistic.
Asterisks indicate statistical significance at the ***1 and **5 percent level.
Independent variable Dependent Variables
1-1 Speech rate 1-2 Articulati on Rate 1-3 ASD 1-4 MLU in syllables 1-5 MLU in seconds 1-6 # Pauses p.m. total 1-7 # Pauses p.m.phon 1-8 Phonation time ratio 1-9 # Filled pauses 1-10 # Dys- fluencies 1-11 Mean Pause Duration FLES -0.257 (1.55) -0.118 (1.02) 0.009 (0.98) -1.483 (1.92) -0.397 (1.84) 0.019 (0.61) 0.145 (2.00) -0.064 (1.51) 0.100 (1.70) -0.007 (0.27) 0.016 (0.05) Fun 0.052 (0.38) 0.161 (0.68) -0.012 (1.57) -0.182 (0.29) -0.108 (0.61) -0.007 (0.26) -0.015 (0.25) -0.006 (0.19) -0.051 (1.06) 0.009 (0.45) 0.361 (1.33) Important 0.339 (1.49) 0.053 (0.34) -0.004 (0.33) 1.545 (1.47) 0.413 (1.40) -0.015 (0.34) -0.107 (1.09) 0.092 (1.60) 0.050 (0.63) -0.042 (1.27) -0.320 (0.71) DVD without subtitles 0.204 (1.25) 0.094 (0.83) -0.007 (0.71) 0.915 (1.47) 0.219 (1.04) -0.006 (0.19) -0.070 (1.00) 0.051 (1.23) -0.034 (0.59) -0.031 (1.31) -0.186 (0.57) Dare to use English 0.048
27 3.7.3. Regression analysis for category home situation
28 Table 3. Regression analysis for the control variable home environment
Number in parentheses are absolute values of t-statistic.
Asterisks indicate statistical significance at the ***1 and **5 percent level
Independent variable Dependent Variable
1-1 Speech rate 1-2 Articulati on Rate 1-3 ASD 1-4 MLU in syllables 1-5 MLU in seconds 1-6 # Pauses p.m. total 1-7 # Pauses p.m.phon 1-8 Phonation time ratio 1-9 # Filled pauses 1-10 # Dis- fluencies 1-11 Mean Pause Duration FLES -0.104 (0.48) -0.033 (0.28) 0.002 (0.16) -0.669 (0.84) -0.201 (0.94) 0.007 (0.22) 0.089 (1.13) -0.033 (0.59) 0.094 (1.76) -0.013 (0.60) -0.175 (0.34) Mixture of languages as mother tongue -0.183 (0.32) -0.002 (0.01) -0.002 (0.06) -0.090 (0.04) -0.057 (0.10) -0.026 (0.29) -0.010 (0.05) -0.051 (0.35) -0.044 (0.31) -0.032 (0.57) 0.170 (0.12) Foreign language as mother tongue 0.328 (0.57) 0.088 (0.28) -0.005 (0.18) -1.116 (0.55) -0.447 (0.79) 0.045 (0.51) -0.093 (0.45) 0.071 (0.49) 0.008 (0.06) 0.048 (0.84) -0.644 (0.47) Mixture of languages spoken at home 0.941 (0.96) 0.825 (1.55) -0.072 (1.60) 7.932 (2.22)** 1.787 (1.85) -0.156 (1.04) -0.263 (0.74) 0.128 (0.51) -0.241 (1.00) -0.078 (0.81) 0.191 (0.08) Foreign language spoken
at home -0.208 (0.41) -0.073 (0.26) 0.005 (0.22) 0.109 (0.06) 0.071 (0.14) -0.017 (0.22) 0.133 (0.71) -0.054 (0.41) 0.075 (0.59) 0.051 (1.00) 0.328 (0.27) English as second spoken
language at home 0.038 (0.16) -0.077 (0.60) 0.005 (0.47) -0.116 (0.14) -0.002 (0.01) -0.002 (0.05) -0.018 (0.21) 0.020 (0.34) 0.010 (0.17) 0.004 (0.19) 0.274 (0.49) Other language as
second spoken language at home -0.256 (0.60) -0.352 (1.50) 0.029 (1.48) 2.524 (1.60) 0.925 (2.18)** -0.081 (1.22) -0.083 (0.53) -0.022 (0.20) -0.071 (0.67) -0.051 (1.20) 0.505 (0.49) Dutch and English as
second spoken languages at home 0.172 (0.45) -0.243 (1.16) 0.025 (1.40) 0.374 (0.27) 0.157 (0.41) 0.008 (0.14) -0.069 (0.50) 0.074 (0.75) 0.125 (1.31) -0.043 (1.13) -0.467 (0.51) Dutch + other language
29 3.7.4. Regression Analysis exposure outside school
As a last step, the regression analysis was conducted to test whether the degree of fluency can be influenced by the exposure of English outside school. The first regression illustrates that the more students write emails, the higher their speech rate is. Writing emails also positively influences the articulation rate, as is demonstrated by regression 1-2. This regression also shows that the more students play computer games, the higher their articulation rate is. In the third regression writing emails and gaming are also significant, yet negatively. Here, the opposite holds: the more students write English emails and the more they game, the lower their mean syllable duration is.
Regression 1-4 demonstrates three significant variables. The regression shows a negative relationship between watching movies with subtitles and the mean length of utterances in syllables. The same holds for the mean length of utterance in seconds (regression 1-5), the more movies without subtitles the students watch, the lower their mean length of utterances in seconds. Regression 1-4 additionally illustrates that writing an email in English and listening to English songs are positively related to the mean length of utterances in syllables. Writing an email and listening to English songs are also positively related to the mean length of utterance in seconds (regression 1-5).
When looking at regression 1-7, the regression proves that non FLES students have a higher number of pauses per minute speaking time than FLES students. The regression furthermore shows a negative influence of writing emails and listening songs on the number of pauses per minute speaking time. In other words: the more participants write emails and listen to songs, the less pauses per minute speaking time they produce. Lastly, regression 1-7 also shows that watching movies with subtitles is strongly significant, yet it has the total opposite effect. The more students watch movies with subtitles, the more pauses per minute speaking time can be counted.
Regression 1-8 almost shows the same significant variables as regression 1-7. However, the influences vary. First, it can be noticed that due to a negative influence of FLES, it can be stated that FLES students have a higher phonation time ratio than non FLES students. Second, reading magazines/books and writing emails positively contribute to the phonation time ratio, whereas watching movies with subtitles results in a lower phonation time ratio.
30
Independent Variable Dependent Variable
1-1 Speech rate 1-2 Articulatio n Rate 1-3 ASD 1-4 MLU in syllables 1-5 MLU in seconds 1-6 # Pauses p.m. total 1-7 # Pauses p.m.phon 1-8 Phonation time ratio 1-9 # Filled pauses 1-10 # Dys- fluencies 1-11 Mean Pause Duration FLES -0.402 (1.97) -0.110 (0.93) 0.011 (1.08) -1.007 (1.65) -0.248 (1.69) -0.001 (0.02) 0.201 (3.21)*** -0.105 (2.05)** 0.080 (1.32) -0.013 (0.49) 0.221 (0.34) Magazine/book 0.343 (2.41) 0.056 (0.68) -0.006 (0.82) 0.726 (1.70) 0.166 (1.62) 0.032 (1.09) -0.079 (1.80) 0.095 (2.66)** 0.021 (0.48) 0.017 (0.89) -0.703 (1.56) Email 0.398 (2.89)** 0.242 (3.05)** -0.022 (3.12)*** 1.325 (3.22)*** 0.281 (2.84)*** 0.003 (0.11) -0.132 (3.13)*** 0.082 (2.38)** -0.072 (1.76) -0.032 (1.78) -0.373 (0.86) Letter/postcard 0.093 (0.63) -0.078 (0.91) 0.009 (1.16) 0.413 (0.93) 0.135 (1.26) 0.002 (0.07) -0.045 (0.98) 0.035 (0.95) 0.039 (0.88) -0.015 (0.77) -0.212 (0.45) Holiday in English speaking Country 0.042 (0.50) 0.027 (0.55) -0.002 0.07 -0.114 (0.45) -0.044 (0.73) 0.007 (0.40) 0.011 (0.42) 0.008 (0.40) 0.015 (0.60) -0.002 (0.14) -0.132 (0.49) Speaking -0.034 (0.20) 0.017 (0.17) -0.000 (0.07) 0.107 (0.21) 0.034 (0.27) -0.007 (0.21) 0.002 (0.04) -0.012 (0.28) -0.086 (1.69) 0.011 (0.50) 0.175 (0.32) Videos internet -0.141 (1.32) -0.075 (1.22) 0.007 (1.35) -0.080 (0.25) 0.013 (0.17) -0.030 (1.41) -0.028 (0.85) -0.029 (1.10) 0.012 (0.37) -0.004 (0.31) 0.314 (0.93) Games 0.164 (1.62) 0.120 (2.06)** -0.011 (2.16)** 0.264 (0.88) 0.037 (0.51) 0.009 (0.45) -0.044 (1.41) 0.032 (1.26) -0.012 (0.39) 0.001 (0.09) -0.258 (0.81) Magazine/book -0.076 (0.55) -0.049 (0.62) 0.005 (0.66) 0.171 (0.42) 0.049 (0.50) -0.036 (1.30) -0.010 (0.23) -0.021 (0.61) 0.077 (1.89) -0.029 (1.62) 0.095 (0.22) Movie + subtitles -0.199 (1.64) 0.014 (0.20) -0.001 (0.14) -1.292 (3.57)*** -0.373 (4.28)*** -0.036 (1.30) 0.176 (4.74)*** -0.063 (2.07)** 0.037 (1.01) 0.010 (0.60) 0.175 (0.46) Songs 0.176 (0.90) 0.139 (1.24) -0.013 (1.34) 1.282 (2.21)** 0.309 (2.21)** -0.057 (1.44) -0.164 (2.75)*** 0.029 (0.60) 0.055 (0.94) -0.040 (1.55) 0.276 (0.45) Info at website 0.168 (1.36) 0.084 (1.17) -0.008 (1.32) 0.413 (1.12) 0.095 (1.07) 0.002 (0.10) -0.064 (1.68) 0.037 (1.20) -0.061 (1.65) 0.011 (0.67) -0.061 (0.16) TV + subtitles -0.015 (0.15) -0.068 (1.21) 0.005 (1.03) -0.164 (0.56) -0.009 (0.12) 0.020 (1.02) 0.019 (0.63) 0.011 (0.47) -0.023 (0.79) 0.019 (1.45) 0.023 (0.08) No. of observations 48 48 48 48 48 48 48 48 48 48 48 R2 0.602 0.469 0.477 0.632 0.645 0.243 0.709 0.594 0.427 0.318 0.250
Number in parentheses are absolute values of t-statistic.
Asterisks indicate statistical significance at the ***1 and **5 percent level.
