Eye Movements and Linguistic Landscape

(1)

Eye Movements and Linguistic Landscape

Pouran Seifi

S2624842

MA in Multilingualism

Faculty of Liberal Arts

University of Groningen

Supervisors:

Dr. Hanneke Loerts

Professor Kees de Bot

(2)

I could never manage to write my thesis without the guidance of my supervisors. Foremost, I would like to express my deepest gratitude to my supervisors, Pro. Dr. Kees de Bot and Dr. Hanneke Loerts for introducing me this research topic. My sincere thanks go to Hanneke for her excellent guidance and patience.

I really appreciate support and best wishes of my parents, sisters and brothers.

I would like to thank my classmate, Nikki Webster, for her help to distinguish different languages in the linguistic landscape of Leeuwarden.

(5)

Abstract

A wealth of study has investigated linguistic landscape for a variety of purposes in different fields such as multilingualism, minority languages, language policy, identity awareness, and language change. However, none of those prominent studies have taken into account objectively the view of people on linguistic landscape while they pass by a street. The aim of the present study was to combine the study of eye movements with linguistic landscape using an eye tracker instrument. Some short videos recorded and sixty-one images taken from Nieuwestad and Waagplein Streets in Leeuwarden, the Netherland were applied as stimuli. Forty-four students from Groningen University participated in this experiment. The eye movements of the participants were recorded and eye tracker metrics of average of fixation count and total fixation duration on signs containing linguistic messages in both images and movies were computed. Two-way ANOVA and one-way MANOVA were calculated to illustrate the effect of native language and gender of the participants on the defined fixation metrics. Proportions of fixation metrics were computed to reveal which linguistic signs and languages on them hold the attention of the viewers. The findings suggest that signs of ad (poster on shop window), stone inscription, a-board, shop name on window, and shop front signs were more attractive for the viewers. Moreover, Dutch and/or English in combination with other(s) were the most attractive languages on the signs. However other foreign languages obtained the least attention. Furthermore, Frisian language hold average attention. Finally, neither native language nor gender of the participants had a considerable effect on fixation metrics of eye movements in the areas of interests on images and videos.

(6)

Chapter 1. Introduction

Introduction

Writings on signs have a major informative and persuasive role in the landscape of cities and villages. Moreover, competition of signs on a-boards, shop windows, and billboards for persuading the people passing by is very prevalent. The different languages on commercial signs reveals the existence of different (minority) language speakers in that region as well. Edelman (2010) states that based on the principle of presentation of self, actors in social life try to obtain their aims by expressing their identities through their linguistic choices. In the linguistic landscape, signs compete with one another to catch the attention of passers-by. Trying to win this struggle, actors present advantageous images of themselves and show their uniqueness. As linguistic landscape actors try to influence the public, they adapt to the values of the audience and focus on the expected attractiveness of the signs.

A wealth of research has investigated linguistic landscapes for a variety of purposes including multilingualism, minority languages, language policy, identity awareness, language changing, and language and power issues. Some of these studies investigated people’s perception and attitudes regarding a language in a place.

Moriarty (2013) argues; however, that traditional approaches to linguistic landscape studies have mostly concentrated on issues of counting languages so as to uncover the hierarchy of linguistic resources in multilingual settings.

Interestingly, Duchowski (2007) claims that if someone’s eye movements is tracked, we can understand the track of attention that the observer applies. This may help us learn what is attractive for the observer. Moreover, it may provide us information that helps us find out how she or he thinks about the scene which that person is viewing.

Statement of purpose

(7)

Chapter 2: Background

The aim of this chapter is to present an overview of previous studies in the field of linguistic landscape and eye movements, and the eye tracking methodology. Before examining and discussing the studies in detain; however, the next section will define the focus of this study, linguistic signs and linguistic landscape.

2.1. Signs and Linguistic Landscape

According to Backhaus’ definition a sign is “any piece of written text within a spatially definable frame including anything from the small handwritten sticker attached to a lamp-post to huge commercial billboards” (2007, P. 66). “Public signs are a type of semiotic sign in that they stand for something other than

themselves. Thus, a public sign is a signifier that relates a company, a product, a place, a rule, or some other concept. On the other hand, a sign indicates a direction on how to get to a place, as in the case of guidance signs, or simply call attention to it, as advertisement signs do”. (p. 5)

Landry and Bourhis (1997) describe linguistic landscape like all linguistic tokens which show the public activities that include road signs, names of sites, streets, buildings, places and institutions as well as advertising billboards, commercials and also personal visit cards. Schiffman (1996) states that large cosmopolitan urban centres are often culturally and linguistically different, composed of diverse and identifiable neighbourhoods, each with its own linguistic culture, that is “the set of behaviours, assumptions cultural forms, prejudices, folk belief systems, attitudes, stereotypes, ways of thinking about language, and religion historical circumstances associated with a particular language” (p. 5).

(8)

such as “public institutions, associations, firms, and individuals that stem from most diverse strata and milieus” (p. 8).

Linguistic landscape actors provide the linguistic elements which are divided into two categories. First, they are used by institutions work under the control of local or national authority. Second, these elements are applied by people or organizations that work independently but within accepted rules (Ben-Rafael et al, 2006). The important difference between aforementioned elements is that the first one tries to show loyalty to the powerful culture in the society, but in the second type, personal desires create linguistic elements (Ben-Rafael et al, 2006). Additionally, Ben-Rafael et al. (2006) divides these linguistic landscape elements into 2 sections. They call the first type top-down and the second one as bottom-up elements.

Backhaus (2007) emphasizes the importance of analysis of the languages which are present on signage in public space, especially in multilingual environments, since “It can provide valuable insights into the linguistic situation of a given place, including common patterns of language and script use, official language policies, prevalent language attitudes, power relations between different linguistic groups, and the long term consequences of language and script contact, among others”(p. 11). Ben-Rafael et al. (2006) state that from the good-reasons perspective, it is possible to explain the features of linguistic landscape by comparing its actors’ desires with the public opinion since the eventual aim of the signs of linguistic landscape is

persuading people and influencing customers. Consequently, both elements present themselves to the people who pass-by, understand, and interpret the signs that they see in the street.

Tufi and Blackwood (2010) argue that since Landry and Bourhis first applied the term in 1997, the linguistic landscape has developed in a number of different

(9)

2.2. Empirical studies on linguistic Landscape

A small selection of prominent linguistic landscape studies are presented in this section.

In order to explain the questions of language contact, language mixing and language dominance Huebner (2006) examined the linguistics landscape of Bangkok neighbourhoods. He highlighted the importance and influence of English as a global language and revealed the extent of linguistic diversity in a language metropolitan area as Bangkok. He claimed that English has an important effect on linguistic landscape of Bangkok. His study revealed the shift the shift from Chinese to English since it is the major language of communication. He also showed that English helps Thai language to develop in the areas of syntax, pronunciation, and spelling.

Backhaus (2005) dealt with multilingual signs in Tokyo, Japan. He studied the distinction between official and nonofficial multilingual signs and power relationship among different linguistic groups. Results of his study revealed that official signs are applied to show existing power relations. However, non-official signs are used for solidarity with non-Japanese issues.

Comparing patterns of linguistic landscape in different homogeneous and mixed Israeli cities, and in East Jerusalem, Ben-Rafael et al. (2006) focused on the degree of visibility of private and public signs of these main languages of Israel-Hebrew, Arabic, and English. They pointed to the importance of linguistic landscape as a means of studying social reality. Their study demonstrated different patterns of linguistic landscape in different communities of Israel. As a result, they state that the signs of Hebrew-English, Arabic-Hebrew, and Arabic-English exist in the

communities of Jewish, Israel-Palestinian, and East-Jerusalem, respectively. Cenoz and Gorter (2006) focused on linguistic landscape of two streets in two multilingual cities in Donostia-San Sebastian and Ljouwert-Leeuwarden where minority languages of Basque and Frisian are spoken. They compared the use of these two languages as related to the differences in language policy regarding the minority language in these two settings. Consequently, they summarized the order of 3

(10)

Seifi (2014) studied linguistics landscape of a street in Leeuwarden and analyzed the use of a minority language (Frisian), the state language (Dutch), English as an international language and other minority languages. Moreover, she compared the proportion of these languages in two different times with referring to the study of Cenoz and Gorter (2006). She showed that Dutch language is the omnipresent in the linguistic landscape in Leeuwarden. The process of globalization is obvious through the presence of English; however, the process of regionalization of Frisian language is not successful.

However, some studies of linguistic landscape concentrated on the attitudes of inhabitants of a region toward the languages on the linguistics landscape. For instance, Cenoz and Gorter (2009) examined the perceptions of languages on signs and stated (Spanish or Basque) language preference of the local inhabitants of the city of Donostia- San Sebastian. They employed environmental economic approach to study linguistic. Outcome of their study show that Basque speakers like to rank the importance of signs in the city as they prefer Basque to be higher than Spanish; however, Spanish speaker are not worried about the status of Spanish since they believe that it is the dominant language at any situation.

Trumper- Hecht (2010) analyzed the diverse perceptions, preferences and attitudes of Arab and Jewish residents toward two official languages of Arabic and Hebrew in Upper-Nazareth. She states that there is a significant difference between perceptions of Arabs and Jews regarding linguistic landscape in the city. Jews believe that Hebrew is the dominant language in linguistic landscape, but Arabs Arabic does not exist. However, although Arab residents know that Hebrew is dominant language in Upper Nazareth, they believe that Arabic signs exist more than Hebrew in this area. It seems that both groups have different cognitive map of linguistic landscape of Upper

Nazareth in their minds.

Additionally, based on a survey analysis about attitudes of students at NHL university of applied sciences toward Frisian language, Flora (2014) indicates that Frisian students and those who live in Friesland presented positive attitudes, while non- Frisians and short stays or those people who had never been here before showed negative attitudes.

(11)

Interestingly, examining language shift from German-Hungarian bilingualism to the exclusive use of German in the community of Oberwart in Austria, Gal (1978) describes how the women of a Hungarian-German bilingual town called Oberwart have contributed to a shift in patterns of language choice. The whole community gradually and systematically changed from stable bilingualism to the use of only one language in their interactions. The differences related to sex in language preference have affected on the whole wide process of change in the community. She argued convincingly how the negative association of Hungarians with peasant-ness led to more young women choosing German as their language of use in a number of social domains. She adds that “women’s speech choice must be explained within the context of their social position, their strategic life choice and the symbolic values of the available linguistic alternatives” (p.1).

The next section will provide an outline of eye movements, visual attention, and eye tracking methods.

2.3. Eye Movements

As Schotter and Rayner (2012) elaborate, visual information is obtained with our eyes when light hits the retina and is changed into electric signals which get passed on the brain in order to be interpreted. Significantly, not all areas of the retina have the same acuity, or resolution. The fovea is the region of highest acuity which is located in the center of the retina and extends 2 degrees of visual angle in diameter (see Figure 1). Outside the fovea, in regions called the parafovea and the periphery acuity decreases quickly. For this reason, eyes must be moved so that the fovea fixates the location of that which we aim to process so as to process information most

(12)

Figure 1. Horizontal section of a human right eye, showing the key components, only small portion of image which falls on the fovea can be seen in high resolution

(Epworth, 2014).

These eye movements are named saccades and last almost 25–60 m (milliseconds - relying on the size of the actual movement). Saccades are quick eye movements which are used in moving the fovea to a new location in the visual environment. Saccadic movements are not only voluntary but also reflexive. The movements can be voluntarily done or they can be evoked as a corrective optokinetic or vestibular

measure (Schotter & Rayner, 2012).

By the same token, Shebilske and Fisher (1983) indicate that saccades range in duration from 10 ms to 100 ms, which is an enough short duration to cause the executor efficiently blind during this transition. Between the saccades, “our eyes are relatively stable (in fixations) which last approximately 200–250 ms. It is during fixations that information is obtained from the visual stimulus since during saccades vision is suppressed and we are essentially blind” (Schotter & Rayner, 2012, p. 84). Accordingly, Duchowski (2007) states fixations are eye movements that stabilize the retina over a stationary object of interest. Fixations are those eye movements that best indicate the locations of the viewer’s overt visual attention.

Rayner (1978, 1998, and 2009) argues that we may have the intuition that we are able to process all the information in a visual display in a single fixation, but it is simply not possible without moving our eyes so that our foveas may fixate a location to process the information therein.

(13)

voluntary and overt visual attention. This assumption includes the reasonable

involuntary utility of these movements, or oppositely, the covert nonuse of these eye movements, for instance, in the case of parafoveal attention. Fixations naturally correspond to the tendency to keep one’s gaze on an area of interest. Similarly, pursuits are used in the same way for objects in smooth motion. Saccades are considered as a clear act of the willingness to voluntarily change the focus of attention.

Additionally, Buswell (1939) believes that eye movements are unconscious small changes to meet the demands of attention while having a visual experience. The most important assumption is that in a visual experience the centre of fixation of the eyes is the centre of attention in a period of time. In this manner, for the present study, it is assumed that fixation eye movements of participants reveals their attention of the signs of linguistic landscape.

2.4. Visual Attention

Duchowski (2007) indicated that visual attention has been investigated for over a hundred years. A good qualitative definition of visual attention was given by the psychologist James (1981), as he believes that attention is known for everybody. It is when mind takes the possession of one thing out of different possible choices or out of a related series of thought that are developing in the mind in a clear and vivid form. The distinctive characteristics of attention are focalization, concentration, and

consciousness. It suggests withdrawal from some things so as to deal effectively with others.

Duchowski (2007) mentions that “humans are finite beings that cannot attend to all things at once. In general, attention is used to focus our mental capacities on

selections of the sensory input so that the mind can successfully process the stimulus of interest” (p. 4). Information processing capacity of human beings is limited. The brain processes sensory input by focusing on specific components of the whole sensory area so that interesting sights, sounds, and smells may be examined with much more attention to detail than peripheral stimuli. This is especially true of vision. Duchowski (2007) believes that “visual scene inspection is performed minutatim, not in toto” (p. 4). It means that perceptions of small areas are collected to build a

(14)

Eye tracking provides the best measure of human attentional behavior (Duchowski, 2007). He emphasizes that “eye movements are particularly interesting because they present measures that can provide insights into the visual, cognitive, and attentional aspects of human performance” (p. 172).

2.5. Eye Tracking Methods

As Duchowski (2007) points out, an eye tracker is the measurement instrument which is most often applied for measuring eye movements. The goal of eye movement measurement and analysis is to gain insight into the viewer’s attentive behavior. In general, “there are two kinds of eye movement monitoring techniques: those that measure the position of the eye relative to the head, and those that measure the orientation of the eye in space, or the ‘point of regard’ (Young & Sheena, 1975, cited in Duchowski, 2007, p. 51 ). Head-mounted systems measure the angles of eyes in the head whereas table mounted systems measure the gaze angles. I am interested in the latter technique in this study.

This section will briefly explain the different methods that can be used to track eye movements. The overview is based on Duchowski (2007)

2.5.1. Electro OculoGraphy

It has been the most widely used eye movement recording method since forty years ago. This method relies on measurements of different potential electric of skin which are recorded by electrodes that are placed around eyes. This technique measures eye movements by the help of a head tracker since this technique measures eye

movements relative to head position. 2.5.2. Scleral Contact Lens/Search Coil

This is one of the most correct techniques of eye movement measurements. In this method an optical object is fixed firmly on a contact lens, then it is worn on the eye. Additionally, a mounting stalk is attached to the lens. This technique measures eye position relative to the head. Therefore, it is not suitable for the measurement of positions of the eye in space.

2.5.3. Photo-OculoGraphy or Video-OculoGraphy

(15)

movements on a videotape. Manual visual inspection can be extremely tiring and likely to be in error.

2.5.4. Video-Based Combined Pupil/Corneal Reflection

None of the aforementioned techniques provide point of regard measurement. Video-based eye trackers apply almost cheap cameras and hardware of image processing in order to calculate the point of regard in real-time. This device can be table-mounted or worn on the head. This is a suitable eye tracking technique in interactive systems. In this technique corneal reflection of the infra-red light relative to the place of the centre of pupil is measured.

The present study used a table- mounted or remote eye tracker so as to measure fixation count and total fixation duration of participants’ eyes to the defined areas of interests while they were viewing movies and images stimuli.

However, it should be noted that mobile eye trackers are a new generation of eye trackers. They are worn as glasses and the persons wearing them are free to do whatever they want. They give researchers the ability to capture truly objective and deep insights into human behavior in any real-world environment. It shows exactly what a person is looking at in real time. This instrument can be used in different fields of study such as psychology, social interactions, group dynamics, control panels, sinage, navigation, mobile device testing, out of home advertising, TV advertising, driving, and in-store shopping. (Mobile eye tracking, 2015)

In the following section the place in which stimuli were recorded, will be explained. Friesland, Leeuwarden

(16)

Figure 2. Map of Leeuwarden Friesland, the Netherlands (Google Maps, 2015).

Dutch is the language spoken in the Netherlands and it is also the official language of the country. However, the second official language of Friesland is Frisian. Most Dutch people speak at least one foreign language, mostly English which is taught at school during the basic education. Many Dutch people also speak German, which is in many aspects similar to Dutch language and some of them speak French. People from the Muslim community often speak Turkish or Arabic (Population and languages of Leeuwarden, 2015).

Interestingly, Bijvoet (2015) argues that Frisian is a language with a distinguished history that parallels that of neighboring Germanic languages like Dutch, English and German. Frisian is closely related to all of them. Claims that Frisian is the closest relative of English are usually supported anecdotally by people recognizing Frisian words which have more in common with English than with Dutch.

Research Questions and Hypotheses

The following questions will be addressed during this research:

Q1. Which signs of linguistic landscape (road signs, billboards, place names, shop signs and public signs) are more attractive for the viewers?

Q2. Which languages of linguistic landscape (Dutch, English, or other) are more attractive for the viewers?

(17)

Q4. Is there any difference between gender of the participants regarding fixation count and total fixation duration in the areas of interests?

Based on the results of studies presented in this chapter, the following hypothesis will be investigated:

H1a: There is a significant effect of native language of the participants on fixation count and total fixation duration in the areas of interests?

It is expected that Dutch speakers may have more fixations on Dutch signs and non- Dutch participants may have more fixations on non-Dutch signs.

H1 b: There is some differences between gender of the participants regarding fixation count and total fixation duration in the areas of interests?

Another expectation is that female and male participants may have different fixation count and duration on different linguistic landscape signs, since the study of Gal (1978) revealed that females may have different language preferences. Females may look at Dutch and English signs with large number of fixation counts and long duration of fixations because Dutch is the national language and English is an international language.

(18)

Chapter 3: Method

The methodology used in the present study contained the following steps:

Determination of locations for recording, recording movies and taking pictures from the selected linguistic landscapes, creating 2 projects containing films and images in Clear View, preparing a questionnaire, naming the type of signs, coding signs based on their languages, creating areas of interests in both projects, selecting participants, recording eye movements of participants while watching those movies and looking at pictures, and visualizing data.

3.1. Participants

Forty-four Groningen University students took part in this experiment. Data were collected from native speakers of Dutch (N= 31, Mean age= 24.8, Range= 19- 44) and from Non-Dutch participants from Japan, England, Morocco, Italy, Basque, Germany, and Wales (N= 13, Mean age= 23, range= 19-36). Seventy five percent of the

participants were born in the Netherlands and 22.2% were born in other countries. In total, thirty-five students were female and 9 of them were male. All subjects had normal or corrected-to- normal vision. Thirty - eight of them will complete their Bachelor’s degree soon and 6 participants are doing a Master. All participants were naïve with respect to the purpose of this study and participated in this experiment in return for course credits.

3.2. Material

3.2.1. Static Stimuli

Using HUAWEI Ascend P6, sixty- one pictures were taken of linguistic landscape of some parts of Nieuwestad Street and Waagplein Streets in Leeuwarden, Friesland, in the Netherlands as illustrated on the map (Figure. 3). The nature of scenes included any front shop with any type of signs on or in front of it and any linguistic sign in that street. Participants were told to look at the pictures while they were on the screen, and no further instruction was provided.

(19)

middle lasting for 2 seconds in order to centre subjects’ gaze position. The images were displayed until the set time duration of 5 seconds expired.

3.2.2. Dynamic stimuli

The movie stimulus was a 10 minutes and 27 seconds colorful video clip. It was recorded using a Canon LEGRIA HF R 506 camera in some parts of Nieuwestad and Waagplein Streets in Leeuwarden, Friesland, in the Netherlands as shown on the map below (Figure. 3). Then, it was split into 3 pieces using Microsoft Movie Maker program, since Tobii Studio eye tracker stopped playing because of lots of areas of interests created on the movie. Then, they were converted into AVI format using Free Make Video Convertor. The order in which the 3 clips were viewed was like the original movie before getting split. The duration of movie elements presentation was set to the end of the video playback.

Figure 3. Map of the areas of Nieuwestad and Waagplein Streets where stimuli were recorded (Google Maps, 2015).

3. 2.3 Defining Areas of Interests in Static and Dynamic Stimuli

(20)

Figure 4. Linguistic landscape signs of hanging banner, shop front sign, ad (Poster on shop window), and flag.

Figure 5. Linguistic landscape signs of a-board, shop canopy, and shop front sign.

The languages on the signs of images were categorized as: Dutch, English, Frisian, French, German, (English, Dutch), (English, Other), (French, Other), (English,

(21)

Figure 6. Language of signs: Dutch, French, and English.

Figure 7. Language of signs: English.

(22)

not all of them were visible clearly and this question was more efficiently addressed using the Images project.

Static AOIs were created using polygons and rectangles drawing tools, then they were named, and were grouped based on the defined language category.

3.2.4. Questionnaire

A sociolinguistic questionnaire, adapted from Loerts (2012), containing some general subject-related questions was included in this experiment. It contains

questions related to personal information such as age, gender, place of birth, level of education and subjects’ linguistic history as well as whether they had ever lived in other countries and what languages they had acquired before starting school (see Appendix 1).

3.3. Apparatus

The research utilized an eye tracking system which includes a Tobii Studio

Computer (Dell Precision T 5500, windows 7), an E-Prime Computer (Dell Precision T 5500, windows 7), and a Tobii T60 eye tracker.

3.4. Procedures

Two projects containing movies and images were created in Tobii Studio program in order to design the eye tracking tests. These projects contain the stimuli to be presented and the participants’ data. Media, images and movies, were added to each test by dragging the media element from the media toolbar to the time line.

To save time, participants were instructed orally. To move from one image element to the next one, view time was selected, 5 seconds for every image was allocated in the view time field.

In movies project, 3 video files were presented to the participants. Duration of the media element presentation was set to the end of the video playback, the next element on the time line was shown when the video ended.

Test preview was the next step. The media elements on the time line in each project were previewed without recording any data in order to verify the correct settings and transitions between the different elements before each recording.

(23)

The next step was setting calibration. The default settings of Tobii Studio for regular calibration was set. Five moving red points on a white background with medium speed on full screen were presented automatically.

Participants were tested individually in the Eye Lab of the Groningen University. They were placed in a chair approximately 64 cm from the eye tracking monitor during the experiment. Before launching the eye-tracking recording, every subject was taken through the calibration procedure. In order to measure characteristics of the subjects’ eyes, calibration procedure was done before starting every recording data in both the dynamic and static media. The track status box was used to adjust the

participants and eye tracking monitor position. The eyes of participants were shown as 2 white dots in the track status box. Calibration started when they were at the centre of the black box, distance indicator showed 50 to 80cm, and color of the bar at the bottom of box was green suggesting the eye tracker accurately captured the participants’ eyes. Then, subjects were instructed to follow the red dot. If calibration was successful, the participant was informed that a movie or images would be seen on the screen within some seconds. Recording was concluded when all the media

(movies and images) had been presented. Between presenting movies and images stimuli every participant was requested to fill a one page questionnaire and served a piece of chocolate to not let them feel bored and get fatigued. Instructing whole procedure of making recordings, filling questionnaire and greeting lasted 30 minutes.

3. 5. Design and Analyses

After recordings were accomplished in both projects, the next step was application of fixation filter to the data to obtain measurements such as fixation count and total fixation duration. I-VT fixation filter was selected and eye selection filter was specified to use the average gaze point calculated from the gaze data samples from both eyes.

However, Data from 12 participants in movies project and 10 participants in images project were discarded because successful calibrations were not achieved. In those cases gaze data samples were less than 50%.

(24)

collectively (see Figure 8 and Figure 9). Bee swarms visualizations were created to illustrate the dynamic distribution of fixations of all participants through all movie elements with coloured dots. Heat maps were created using different colours to show the number of fixations that participants made in specific areas of images and movies (see Figure 10 and Figure 11).

Figure 8. Gaze plot of fixations of all participants on an image. The fixation of each participant is illustrated with a coloured dot.

(25)

Figure 10. Heat map of fixations on the areas of interests on an image. The red colour indicated the highest number of fixations or longest times And purple colour shows the least ones and shortest time of fixations.

(26)

The first step in the analysis of eye movements consisted of calculating eye tracking metrics and statistics. The metrics of fixation count and total fixation duration within the recorded eye movements were identified. Based on the defined AOIs groups in each project, the Tobii Studio Statistics tool was applied to calculate eye tracking metrics’ descriptive statistics. Descriptive data were saved in another computer for further computation. Sum and percentages of fixation counts on every AOIs group were calculated and displayed on a bar chart to address overall attention allocated by participants to the AOIs. Moreover, data were analyzed with SPSS 20 statistical software.

(27)

Chapter 4: Results

This chapter is divided into two sections since there are two main research questions to be addressed, one concerning eye movements’ fixations on the images and the other one related to testing fixations on the movies. Descriptive analysis including means and standard deviations were prepared to investigate the proportion of fixations on any sign within the defined AOI groups in both projects. Furthermore, inductive parametric statistics (one-way MANOVA and two-way ANOVA) were computed to demonstrate the possible effects of the independent variables (native language and gender) on the dependent variables (fixation metrics) to clarify the possible effects and interactions between the independent variables. Gender with 2 levels of female and male, and native language with 2 levels of Dutch and non-Dutch speakers are nominal independent variables in both projects. Interval dependent variables in image project include fixation count and total fixation duration on AOI groups of Dutch, English, Frisian, (English, Dutch), (English, Other), (French, Other), (English, French), (English, Dutch, English), (English, Dutch, French), (French, Dutch), French, German, and Other. Interval dependent variables in the movies project consist of fixation count and total fixation duration on AOI groups of a-board, ad (poster on shop window), bicycle facility, extruded street name, flat blade street name, graffiti, hanging banner, place name, shop canopy, shop front sign, shop name board, shop name on a wall, shop name on window, stone inscription, and traffic control sing.

The set alpha levels is .05 and all tests are two-tailed since no specific direction was hypothesized for any of the effects. However, to reduce the chance of a Type 1 error in one-way MANOVA tests, Bonferroni adjustment was applied to set a higher level of alpha level. Normal alpha level (.05) was divided by the number of depended variables. Therefore, alpha level in MANOVA computations is .003.

(28)

4.1. Analysis of Images Project

4.1.1. Fixation Count on AOI groups

As illustrated in Figure 12, Dutch and non-Dutch participants made the highest proportion of fixation counts on the following AOI groups on the images: Dutch (35%, 34.5%), Dutch English (18%, 22%), English (15%, 15%). However, they had the least mean number of fixations on English Dutch French (0.7%, 0.6%) and English French Other (0.4%, 0.4%) AOI groups. Additionally, Dutch and non-Dutch participants made (0.4%, 0.4%) of proportion of fixation count in Frisian language on the signs.

Figure 12. Proportion of fixation count of Dutch and non-Dutch participants in AOI groups on the images.

Figure 13 indicates that female and male participants had the highest proportion of fixation counts in Dutch (35%, 35.5%), Dutch English (18%, 18%), and English (15%, 15%) AOI groups on the images; however, English French Other (0.4%, 0.4%) and English Dutch French (0.7%, 0.6%) received the least fixation count from

(29)

Figure 13. Proportion of fixation count of female and male participants in AOI groups on the images.

4.1.2 Total Fixation Duration in AOI groups

(30)

Figure 14. Proportion of total fixation duration of Dutch and non-Dutch participants in AOI groups on the images.

(31)

Figure 15. Proportion of total fixation duration of female and male participants in AOI groups on the images.

4.1.3. Analysis of Average of Fixation Counts on Overall AOI groups on the Images A two-way ANOVA was run to determine if gender and native language variables had any effect on the number of fixations in the overall AOI groups on the images. The average of fixation count for each level of gender and native language variables were normally distributed, as assessed by Shapiro-Wilk test (p > .05), homogeneity of variances was satisfied, assessed by Levine’s test for equality of variances (p >.05).

Statistically, the results revealed no significant effect for gender variable, F (1, 30) = 1.44, p = 0.23; and no significant effect for the native language variable, F (1, 30) = 0.23, p = 0.63. Additionally, the interaction between gender and native language was not significant, F < 1.

(32)

AOI groups on the images. Dependent variables of average of fixation count on each AOI groups (Dutch, Dutch English, English, English Dutch French, English French, English French Other, English Other, French, French Dutch, French Other, Frisian, German, and Other) were used. Result of Levene’s test of equality of error variance indicates that homogeneity of variance was not violated. Statistically, there was no significant difference between Dutch and non-Dutch participants on the combined dependent variables. F (13, 20) = 1.13, p = .3; Pillai’s trace = .42; Partial eta squared = .42.

Additionally, a one-way between groups multivariate analysis of variance (MANOVA) was performed to investigate gender differences in the average of fixation count in AOI groups on the images. Dependent variables of average of fixation count on each AOI groups (Dutch, Dutch English, English, English Dutch French, English French, English French Other, English Other, French, French Dutch, French Other, Frisian, German, and Other) were used. Result of Levene’s test of equality of error variance shows that homogeneity of variance was not violated. Statistically, there was no significant difference between female and male participants on the combined dependent variables. F (13, 20) = 1.48, p = .2; Pillai’s trace = .49; Partial eta squared = .49.

4.1.5. Analysis of Average of Total Fixation Duration in Overall AOI groups on the Images

Results of test of homogeneity of variances revealed it is safe to assume that those groups are equal in variances, (p > .05). A two- way ANOVA was run to determine if gender and native language had any effect on average of total fixation duration on overall AOI groups in the images. Results indicated that gender group had no

(33)

4.1.6. Analysis of Average of Total Fixation Duration in Each AOI group on the Images

A one-way between groups multivariate analysis of variance (MANOVA) was performed to investigate native language differences in average of total fixation duration in each AOI groups on the images. Dependent variables of average of total fixation duration on each AOI groups (Dutch, Dutch English, English, English Dutch French, English French, English French Other, English Other, French, French Dutch, French Other, Frisian, German, and Other) were applied. Result of Levene’s test of equality of error variance indicated that homogeneity of variance was not violated. Statistically, there was no significant difference between Dutch and non-Dutch participants on the combined dependent variables. F (13, 19) = 2.21, p = .057; Pillai’s Trace = .42; Partial eta squared = .6.

Furthermore, MANOVA was performed to investigate gender differences in average of total fixation duration in each AOI groups on the images. Dependent variables of average of total fixation duration on each AOI groups (Dutch, Dutch English, English, English Dutch French, English French, English French Other, English Other, French, French Dutch, French Other, Frisian, German, and Other) were used. Result of Levene’s test of equality of error variance showed that homogeneity of variance was not violated. Statistically, there was no significant difference between female and male participants on the combined dependent variables. F (13, 19) = 1.51, p = .2; Pillai’s Trace = .5; Partial eta squared = .5.

4.2. Analysis of Movies Project

4.2.1 Fixation Count on the AOI groups

(34)

Figure16. Proportion of fixation count of female and male participants in AOI groups on the movies.

(35)

Figure 17. Proportion of fixation count of Dutch and non-Dutch participants in AOI groups on the movies.

4.2.2 Total Fixation Duration in AOI Groups on the Movies

(36)

Figure18. Proportion of total fixation duration of Dutch and non-Dutch participants in AOI groups on the movies.

(37)

Figure 19. Proportion of total fixation duration of female and male participants in AOI groups on the movies.

4.2.3. Analysis of Average of Fixation Counts in Overall AOI Groups on Movies Since sample size is less than fifty, Shapiro- Wilk was applied to test normal distribution of female and male sizes, (p > .05). So, it is normally distributed in both gender groups. Test of homogeneity of variances indicates that it can be assumed that two groups of female and male are equal in variances, (p > .05). However, the sample size in two groups of Dutch and non-Dutch speakers are not normally distributed. The number of Dutch participants were more than non-Dutch ones. But, the two groups are equal in variances. Post-hoc cannot be computed because independent variables have not more than two levels.

(38)

4.2.4. Analysis of Average of Fixation Count in Each AOI Group on the Movies. A one-way between groups multivariate analysis of variance (MANOVA) was performed to investigate native language differences in average of fixation count in each AOI group on the movie elements. Dependent variables of average of fixation count on each AOI group (A-board, ad (poster on shop window), bicycle facility, extruded street name, flat blade street name, graffiti, hanging banner, place name, shop canopy, shop front sign, shop name board, shop name on a wall, shop name on window, stone inscription, and traffic control sing) were applied. Result of Levene’s test of equality of error variance indicated that homogeneity of variance was not violated. Statistically, there was no significant difference between Dutch and non-Dutch participants on the combined dependent variables. F (16, 38) = .48, p = .84; and Pillai’s Trace = .72; Partial eta squared= .7.

Moreover, a one-way between groups multivariate analysis of variance

(MANOVA) was conducted to investigate gender differences in average of fixation count in each AOI group on the movie elements. Dependent variables of average of fixation count on each AOI group (A-board, ad (poster on shop window), bicycle facility, extruded street name, flat blade street name, graffiti, hanging banner, place name, shop canopy, shop front sign, shop name board, shop name on a wall, shop name on window, stone inscription, and traffic control sing) were used. Result of Levene’s test of equality of error variance showed that homogeneity of variance was not violated. Statistically, there was no significant difference between female and male participants on the combined dependent variables. F (16, 3) = 15, p = .02; and Pillai’s Trace = .98; Partial eta squared = .9.

Set alpha level = 0.003.

4.2.5. Analysis of Average of total fixation duration on overall AOI groups on the movies

Results of Shapiro-Wilk and Levene’s test revealed that samples in both groups of gender and native language variables are normally distributed, ( p > .05), and it is safe to assume that all groups are equal in variances, ( p > .05).

A two-way ANOVA was run to establish if gender and native language of

participants had an effect on average of total fixation duration of their eye movements in overall AOI groups in movie elements. Results indicated that there was no

(39)

for native language variable, F (1, 30) = 0.1, p = 0.7. Moreover, the interaction between gender and native language was not significant, F = 0.3, p = .56. 4.2.6. Analysis of average of Total fixation Duration on each AOI group on the movies

In order to investigate native language differences in average of total fixation duration in each AOI group on the movie elements, a one-way between groups

multivariate analysis of variance (MANOVA) was performed. Dependent variables of average of fixation count on each AOI group (A-board, ad (poster on shop window), bicycle facility, extruded street name, flat blade street name, graffiti, hanging banner, place name, shop canopy, shop front sign, shop name board, shop name on a wall, shop name on window, stone inscription, and traffic control sing) were applied. Result of Levene’s test of equality of error variance indicated that homogeneity of variance was not violated. Statistically, there was no significant difference between Dutch and non-Dutch participants on the combined dependent variables. F (5, 1) = 1.8, p = .5; Pillai’s Trace = .9; Partial eta squared = .6.

Furthermore, a one-way between groups multivariate analysis of variance (MANOVA) was conducted to investigate gender differences in average of total fixation duration in each AOI group on the movies. Dependent variables of average of total fixation duration on each AOI group (A-board, ad (poster on shop window), bicycle facility, extruded street name, flat blade street name, graffiti, hanging banner, place name, shop canopy, shop front sign, shop name board, shop name on a wall, shop name on window, stone inscription, and traffic control sing) were applied. Result of Levene’s test of equality of error variance showed that homogeneity of variance was not violated. Statistically, there was no significant difference between female and male participants on the combined dependent variables. F (5, 1) = 1.19, p = .5; Pillai’s Trace = .8; Partial eta squared = .8.

Chapter 5: Discussion

(40)

count and total fixation duration on different AOI groups in movie and image projects were recorded. It was assumed that this study could provide an efficient means of discovering which specific signs and which languages on the signs on linguistic landscapes in Leeuwarden contribute to their success or failure in keeping the viewers’ attention. It was hypothesized that native language of the participants has a significant effect on average of fixation count and total fixation duration of eye movements in the areas of interests. Moreover, male and female participants differ in terms of average of fixation count and total fixation duration of eye movements in the areas of interests.

The results which have been demonstrated so far illustrate which signs on the movie elements caught the visual attention of the participants more. Moreover, they described which monolingual, bilingual, and multilingual signs on the image elements were more attractive to the participants.

Groups of areas of interests in movie elements were categorized as a-board, ad (poster on shop window), bicycle facility, extruded street name, flat blade street name, graffiti, hanging banner, place name, shop canopy, shop front sign, shop name board, shop name on a wall, shop name on window, stone inscription, and traffic control sing were studied. However, different areas of interests on images grouped as Dutch, English, Frisian, (English, Dutch), (English, Other), (French, Other), (English, French), (English, Dutch, English), (English, Dutch, French), (French, Dutch), French, German, and Other.

Results showed that overall eye movements’ parameters of average of fixation count and total fixation duration were greater for AOI groups of movie elements in the following signs: ad (poster on shop window), stone inscription, a-board, and shop name on window. On the contrary, sings of shop canopy, flat blade street name, graffiti, and bicycle facility received the least fixation metrics.

(41)

findings in this study may suggest that linguistic signs of ad (poster on shop window), stone inscription, shop front sign, a-board, traffic control sign, and shop name on window are attractive for the viewers, but the signs of shop canopy, flat blade street name, and graffiti are not attractive to them.

Regarding investigation of the preferred languages on the signs of linguistic landscape in Leeuwarden, results indicate that the following monolingual and

bilingual AOI groups in the image elements had the highest average of fixation count and total fixation duration: Dutch, (Dutch, English), English, and (English, Other). However, signs with (English, French, Other), (English, Dutch, French), (French), (French, Other), and minority languages received the least average of fixation count and total fixation duration.

We can appeal to the idea of Duchowski (2007) that by tracking eye movements, we can follow the path of observer’s attention and understand what is interesting for the viewers within linguistic landscape. The results suggest that monolingual signs of Dutch and English were more attractive than other minority languages which were present in the linguistic landscape in this study. Accordingly, considering bilingual signs, results indicate that (Dutch, English) and (English, Other) signs were the most favourite ones.

Furthermore, as far as the average of fixation count and total fixation duration of participants on Frisian language signs were concerned, participants spent average time on that language. It is worth mentioning that Frisian was not among the less attractive signs in linguistic landscape of Leeuwarden.

Interestingly, detailed analysis of data in the present study revealed that native language and gender did not have any effect on the average of fixation metrics on the AOI groups on the image elements.

(42)

more preferred than Frisian. This finding is almost in contrary with the findings of Cenoz and Gorter (2009) in Basque country regarding perceptions and attitudes of passers-by. They stated that in Basque country people prefer Basque language to be ranked higher than Spanish, but in the linguistic landscape of Leeuwarden, viewers prefer Dutch, English, and Frisian, respectively.

This study demonstrates that viewers’ and language actors’ (Ben-Rafael et al, 2006) preferences are in agreement with each other in this region. Viewers have tendency toward English signs not only because it is prestigious international

language but also since all participants can communicate in English fluently. Hence, preference of viewers may depend on the languages that they know very well. I would like to add that if we apply a mobile eye tracker and have more Frisian participants, we can have a precise information on the preference of viewers regarding Frisian language in Friesland.

We may conclude that linguistic landscapes and language actors can apply other minority languages on the sinages. This will be beneficial not only to the actors to persuade the passers-by, but also to help minority language speakers to preserve their languages in the society. In a study Backhaus (2005) showed that non-official signs are used for expressing solidarity with non-Japanese issues. I believe that linguistic landscape actors in Friesland should encourage language actors to apply Frisian language on preferred signs to demonstrate solidarity with Frisians.

Results of proportion of different sinages in linguistic landscape in the present study may have some implications to linguistic landscape actors. In order to make a language visible and help to revitalize any minority language in a society, they can apply that language on the preferred sinages as this study suggestes: ad (poster on shop window), stone inscription, a- board, shop name on window, and shop front signs.

Limitations of the study

(43)

limited to Groningen university students; therefore, it had limited generalizability to a city population as a whole.

Second, this study would have benefited from recording movies and taking pictures in multilingual and multicultural cities like Amsterdam and Utrecht.

Additionally, season may be another alternative to improve the quality of recordings. Winter is not as suitable as spring for recording clear movies.

Third, regarding apparatus, mobile eye tracker instrument for recording eye movements of viewers would be better than remote table mounted eye tracker device. Since it provides researcher an opportunity to record eye movements of participants while they are walking through streets and passing by linguistic landscapes.

(44)

Chapter 6: Conclusion

The purpose of the present thesis was to answer the following main questions: Q1. Which signs of linguistic landscape (road signs, billboards, place names, shop signs,

and public signs) are more attractive for the viewers?

Q2. Which languages of linguistic landscape (Dutch, English, or other) are more attractive for the viewers?

Q3. Is there any significant effect of native language of the participants on average fixation count and total fixation duration in the areas of interests?

Q4. Is there any difference between genders of the participants regarding average fixation count and total fixation duration in the areas of interests?

The findings suggest that signs of ad (poster on shop window), stone inscription, a-board, shop name on window, and shop front signs received the highest number of fixations and total fixation duration, respectively. Hence, those signs were more attractive for viewers. Moreover, monolingual and bilingual signs of (Dutch), (Dutch, English), (English), (English, Other Languages) were the most attractive languages on the signs. However, multilingual signs of (English, Dutch, French) and (English, French, Other), and minority languages of, German, French, Azerbaijani, Albanian, Portuguese, and Swedish obtained the least attention. Furthermore, Frisian language hold average attention.

This study hypothesized that gender and native language have significant effect on the average fixation count and total fixation duration in the areas of interests on image and movie elements. The results indicate that native language and gender of the participants do not have a considerable effect on the average fixation count and total fixation duration.

In sum, native language speakers grouped as Dutch and Non-Dutch and gender grouped as female and male did not mainly influence fixation metrics of eye movements on the defined areas of interests in linguistic landscape signs.

(45)

Suggestions for Further Research

If there was no financial or time restriction, I would prefer to choose different participants with diversity of age, ethnicity, and languages with large sample size. Additionally, I would like to record my data in a metropolitan business centre. Combining study of eye movements and linguistic landscape in the prominent fields of linguistic landscape like language and identity, language and attitude, and language and power in countries that application of minority languages are banned, would be interesting.

Moreover, I suggest that combining studying eye movements of different minority language speakers and Dutch speakers with persuasive appeals of commercial signs in linguistic landscape using eye tracker instrument would be appealing. It can

(46)

Appendices

Appendix 1 Questionnaire A- 1. Personal Information Name: ………. Email address: ……… Sex: Male Female

Date of birth: ……… Place of birth: ………

What is the highest level of education that you will complete? ……… A-2. Sociolinguistic Questions

Questions concerning sociolinguistic background Have you ever lived in other countries?

Yes No

If so, please specify when, where, and how long: ……….……… What language(s) did you acquire before starting school?

Dutch

Dutch and (an) other language(s)

(an) Other language(s): ………

What languages did you acquire at school or at work? And how proficient are you in those languages? Please give a number for each of the four skills:

1 = very bad, 2 = bad, 3 = average, 4 = good, 5 = very good

Language: ………… …reading: …….writing: …….….speaking: …...….listening: …….

Language: ……… ……reading: …….writing: …….….speaking: ……….listening: …….

Language: …… ………reading: …….writing: ……….speaking: …….….listening: …….

(47)

What languages did you acquire outside work/school? And how proficient are you in those languages? Please give a number for each of the four skills:

1 = very bad, 2 = bad, 3 = average, 4 = good, 5 = very good

(48)

References

Backhaus, P. (2005). Signs of multilingualism in Tokyo: A diachronic look at the linguistic landscape. International Journal of the Sociology of Language, 175/176, 103–121. Backaus, P. (2007). Linguistic Landscapes: A comparative study of urban multilingualism

in Tokyo, Clevedon: Multilingual Matters

Ben-Rafael, E, Shohamy, E, Amara, M. H, & Trumper-Hecht, N. (2004). Linguistic Landscape and Multiculturalism: A Jewish–Arab Comparative Study. Tel Aviv: Tami Steinmetz Center for Peace Research.

Ben-Rafael, E. (2009) ‘A Sociological Approach to the Study of Linguistic Landscapes’. In Shahomy, E. & Gorter, D. (eds).

Bijvoet, T. (2015). Language of Friesland, Frisian. The magazine about the Netherlands and its people at home and abroad. Retrieved from

http://www.mokeham.com/dutchthemag/language-eating-lekker-2/.

Buswell, G. T. (1935). How People Look at Pictures. Oxford, England: Chicago: Univ. Chicago Press.

Cenoz, J & Gorter, D. (2006). Linguistic landscape and minority languages. International Journal of Multilingualism, 3 (1), 67–80.

Cenoz, J. & D. Gorter (2009). ‘Language Economy and Linguistic Landscape’ in Shohamy, E. & D. Gorter (eds.) Linguistic Landscape: Expanding the Scenery. Oxon & New York: Routledge.

Duchowski, A. T. (2002). A breadth-first survey of eye-tracking applications. Behavior Research Methods, Instruments, & Computers, 34(4), 455-470.

Duchowski, A. (2007, September 14). Eye tracking methodology: Theory and practice (Vol. 373). Springer Science & Business Media.

Edelman, L. J. (2010). Linguistic landscapes in the Netherlands: a study of multilingualism in Amsterdam and Friesland. LOT.

Epworth, R. (2014). Horizontal section of a human right eye. Bottleneck, our human interface with reality. Retrieved from

http://www.humanbottleneck.com/index/images/.

Facts and statistics about the Netherlands. (2015). Map of the Netherlands. Retrieved from http://www.iamexpat.nl/expat-page/the-netherlands/facts-statistics.

(49)

Gal, S. (1978). Peasant men can't get wives: Language change and sex roles in a bilingual community. Language in society, 7(01), 1-16.

Google Map. (2015). Nieuwestad. Retrieved from

https://www.google.com.eg/maps/@53.2010776,5.7935239,331m/data=!3m1!1e3?hl =en

Hilton, N. H., Gooskens, C., & Van Bezooijen, R. (2011). Attitudes towards Frisian in the Netherlands. Oral presentation held at Poznan Linguistics Meeting.

Huebner, T. (2006). Bangkok’s linguistic landscapes: Environmental print, code mixing, and language change. International Journal of Multilingualism, 3(1), 31–51. James, W. (1981).The Principles of Psychology (Vol. I). Cambridge, MA: Harvard University Press.

Labov, W. (I973). The linguistic consequences of being a lame. Language and Society, 2, 81- 115.

Landry, R & Bourhis, R.Y. (1997). Linguistic landscape and ethnolinguistic vitality: an empirical study. Journal of Language and Social Psychology, 16 (1), 23–49.

Leeuwarden (Municipality, Friesland). (2015). Retrieved

fromhttp://www.citypopulation.de/php/netherlands- admin. php? adm2id=0080

Loerts, H. (2012). Uncommon gender: eyes and brains, native and second language learners, & grammatical gender. University of Groningen.

Moriarty, M. (2012) ‘Language Ideological Debates in the Linguistic Landscape of an Irish Tourist Town’. In Gorter, D., Marten, H.F., and Mensel, L. V. (eds).

Population and Languages of Netherlands. (2015). Retrieved from http://www.amsterdam.info/netherlands/population/.

Rayner, K. (1978). Eye movements in reading and information processing. Psychological bulletin, 85(3), 618.

Rayner, K. (1998). Eye movements in reading and information processing: 20 years of research. Psychological bulletin, 124 (3), 372.

Rayner, K. (2009). Eye movements and attention in reading, scene perception, and visual search. The quarterly journal of experimental psychology, 62 (8), 1457-1506. Schiffman, H. F. (1998). Linguistic culture and language policy. Psychology Press. Schotter, E. & Rayner, K. (2012). Eye Movements in Reading, Implications for Reading

(50)

Shebilske, W. L., & Fisher, D. F. (1983). Eye movements and context effects during reading of extended discourse. Eye movements in reading: Perceptual and language processes, 153-179.

Seifi, P. (2014). Linguistic Landscape and Minority Languages in Leeuwarden. Poster presented in a course meeting, Leeuwarden, NL.

Tobiipro. (2015). Mobile eye tracking, Tobii glasses 2. Retrieved from

http://www.tobii.com/de/eye-tracking-research/global/products/hardware/tobii-glasses-eye-tracker/

Trumper-Hecht, N. (2010). Linguistic landscape in mixed cities in Israel from the perspective of ‘walkers’: The case of Arabic. Linguistic landscape in the city, 235-251.

Eye Movements and Linguistic Landscape