Investigation of the achievement scores of the people learning Turkish as a foreign language according to linguistic distance Turkish Title of Article: Yabancı dil olarak Türkçe öğrenenlerin sınav başarılarının dil bilimsel uzaklığa göre incelenmesi Author(s):

Bu arastirmada basta dil ailesi olmak uzere yas, cinsiyet ve bolge degiskenlerinin, uluslararasi ogrencilerin Turkce kurs bitirme sinavindan aldiklari puanlari yordama duzeyleri coklu regresyon yontemiyle incelenmistir. Calisma grubunu Gazi ve Hacettepe Universitelerinin bunyesinde hizmet veren Turkce Ogretim Merkezlerine (TOMER) kayitli 280 ogrenci olusturmaktadir. Veriler, Turkce kurs bitirme sinav kA¢gitlari ve kisisel bilgi formlarindan elde edilmistir. Bulgulara gore, ogrencilerin ortalama puanlari Hami-Sami, Hint-Avrupa, Bantu, Cin-Tibet ve Avustronezya dil ailelerinde Altay dil ailesine gore daha dusuktur. Ogrencilerin yazma puanlari ise Hami-Sami ve Avustronezya dil ailelerinde; konusma puanlarinin Hami-Sami, Hint-Avrupa, Avustronezya ve Cin-Tibet dil ailelerinde; anlama puanlarinin Hami-Sami, Hint-Avrupa, Bantu ve Cin-Tibet dil ailelerinde; dil bilgisi puanlarinin ise Cin-Tibet ve Avustronezya dil ailelerinde Altay dil ailesine gore daha dusuk oldugu gorulmustur. Ayrica yas degiskeninin konusma puanlari uzerinde pozitif bir etkiye sahip oldugu bulgusuna ulasilirken bolge ve cinsiyet degiskenlerinin puanlarin anlamli bir yordayicisi olmadigi gorulmustur. Bulgular alan yazin isiginda tartisilmis ve ileride yapilacak arastirmalara yonelik onerilerde bulunulmustur.


Introduction
In the world of living beings that surrounds us, language that gives a great advantage to humans over the other creatures is also one of the greatest obstacles in communicating with their own kind.Although physical characteristics differ based on biological taxonomy, most animals belonging to the same genus and family can easily communicate with each other, but for full communication human beings have to share the same language codes with others.While the diversity and complexity of the human language creates a great barrier for communication, humans' activities on earth have led them to know different languages.As in the past, knowing more than one language today and having the ability to communicate with people from different languages keep much of its significance.In the age of knowledge "the prominence of knowing at least one foreign language has become evident" (Gömleksiz & Elaldı, 2011, p.444).
Until recently, foreign language was used as a term to describe languages other than mother tongue, but today the term "foreign language" is used to refer to "language that is learned to communicate with people from other nations, or to read a book written in a foreign language or a scientific writing" (Stern, 1991, p.16).For situations where the term "foreign language" cannot be met, the second language term is being used.The second language is the language that is learned after acquiring mother tongue, at the new environment in which person settles for various reasons."At the same time, the second language also includes the third, fourth, fifth languages learned after the mother tongue" (Ellis, 2008, p.5).The increase in interaction between languages has made it possible for foreign and second language learners to spread globally independent of the geography.Learning a language as a foreign language or as a second language is treated as a subset of cognitive and affective dimensions of human learning.According to Tura (1983, p.15), language learning is closely linked to one's learning strategies and styles.It is also related to learning a second culture.Learning a foreign language involves acquisition of a new linguistic system composed of meanings and sounds; learning different styles of communication functions and conversational rules appropriate and valid for this system.Some factors deeply influence the success of learners' language learning process.These factors include the exposure to the language, the duration and extent of language teaching/learning, teacher features, the choice of teaching method and the strategy, textbooks and equipment, environment and the number of students.Besides, the individual characteristics and previous experiences of the person make language learning different for each person.For this reason, "some characteristics of the student should be known in order to ensure that the learning takes place in the most effective way" (Aktaş, 2012, p.30).There are many views on the definition of the individual differences that language learners have and the way in which the learning process is designed accordingly (Ellis, 2003;Robinson, 2002;Skehan, 1989).However, there is no consensus regarding which individual characteristics of the person affect the language learning process.There are two basic approaches to classify individual differences in language teaching: hierarchical and sequential.The starting point of the hierarchical approach is how individual differences lead to learning.This approach tends to treat the concept of individual difference as a whole.The sequencing approach tries to identify individual differences and reveal the relationship of these properties to each other.Ellis (2003), lists them as susceptibility, learning style, learning strategy, personality, motivation, anxiety, volunteerism and belief.Chastain (1988) examines individual differences under four headings: affective, cognitive, social, and biological.According to him, affective differences include personality, attitude, effort, interests and needs; cognitive differences cover knowledge of experiences, learning skills and strategies, tendency and intelligence; social differences refer to social context, language and culture; and biological differences are composed of sex and age.Similar to this view, Horwitz (2008) suggests that affective, cognitive, and metacognitive differences constitute the individual character in language learning.In addition to motivation, attitudes, tendencies, cognitive strategies, age, personality, intelligence, gender, empathy, for individual differences of language learners, Cook (2001) also adds the person's mother tongue and other languages in his or her repertoire.
Whether the mother tongue or other languages that the person knows are influential in the learning of a new language is a common research topic in educational science and linguistics.Each language contains a system of thought.As the person learns a new language, he/she expects resemblance between the patterns of his own language and the new language.The topics of why some people with certain characteristics learn a certain language easier than others, or why a person can learn language A more easily than language B, and the effect of the similarities between languages on language leaning have been investigated by researchers (Dewale, 1998;Hammarberg, 1998;Ringbom, 1987).
Language typology research, which examines similarities and differences between languages, classifies languages, and reveals universal rules about language structure, helps people understand how they use both new language features and language features that they already know when they learn a new language.According to Uzun (2013) some earlier prejudices about the ways that language typology uses to explain such features have gained a considerable prevalence in language teaching environments: some languages are easier or more difficult to learn than others; if foreign language is similar to the mother tongue, it facilitates the learning of the new language.
The similarity between languages is sometimes hereditary, sometimes coincidental.Apart from these, there are similarities between languages that have interacted with each other in the historical process.According to the Ethnologue database, there are 152 separate language families in the world.A language that is a member of any of these language families inherits similarities to other members of the family.Languages that belong to the same language family have close word parallels, linguistic knowledge, syntax and sound features.This can affect the individual's performance to learn the language.The American Foreign Relations Service groups languages that can be learned easily by English speakers in four groups.According to this grouping, it is easier for a person who speaks English to learn the same root in English as Danish, Dutch and French, and in the first group of the same continental circle, and in the fourth group of different roots such as Arabic, Chinese and Japanese.
The similarities and the differences that a language has with another language are expressed by the concept of linguistic distance.The concept of linguistic distance creates a framework for grammatical rules, meanings of the words, alphabets and writing rules, idioms and phrases.Crystal (1987, p.371), defines linguistic distance as "one of the factors that can affect foreign language learning of the structural similarities between languages".Corder (1978) associates the assumption that the linguistic distance in language teaching is more advantageous in situations where there are a lot of similarities between the foreign language (the learned language) and the mother tongue compared to when there are few similarities.In the research carried out by Ringbom (1987), a group of students whose mother tongues were Finnish and whose second languages were Swedish were asked to write English texts.It has been seen that students use the meaning and function of Swedish words more in English text production.In addition, Finnish as the mother tongue is determined to be a less beneficial language.At the end of the study, it was revealed that the English language learning process was more influenced by Swedish language, which is grouped in the same language family as English, than by students' mother tongue Finnish.Cenoz (2001) found in her study on English learners who speak Spanish and Basque as a mother language that linguistic distance is also influential in English learning like age, purpose and competence and that learners transferred more from Spanish, which is linguistically closer to English than Basque language.In their research on Jewish immigrants in Israel, Beenstock, Chiswick and Repetto (2001) pointed out that those who speak Arabic have a higher level of competence in Hebrew than other immigrant groups, which is linked to the linguistic distance between Arabic and Hebrew.Elder and Davies (1998) found that the relationship between the success of the English certificate exam results in Australia and the candidates' native languages does not fully support Corder's (1978) hypothesis that the inter-language familiarity with language distance will accelerate language learning.According to the results of the research, linguistic distance should be taken as an element considered together with other variables rather than being a variable affecting language learning alone.
If the linguistic distance is handled in the context of Turkish, it can be said that Turkish, as an Altaic language interacting with many languages in the historical process, has common points with most languages in terms of the origin, grammar, meaning and utterances.Turkish, like Mongolian, has similar origins with the languages of Altaic language family.Apart from this, they are separated from each other for various reasons and have deep ties with Turkish dialects that have formed their own written language forms today.It can be said that there are common points with languages such as Serbian, Hungarian, Greek, Russian, Bulgarian, Armenian as the languages belonging to different language families but interacting with Turkish (Karaağaç, 2009;Özkan & Musa, 2004).Lots of words have been transferred into Turkish language from Arabic, which is an Afro-Asiatic language, and Persian and French as Indian-European languages.Eker (2011) points out that the terms and proverbs in Turkish, Arabic and Persian have similar meanings.This indicates that there is not only word-level exchange with already mentioned languages, but they also have similarities in terms of meaning.
The familiarity between languages and cultures continues in various fields today.Turkmenistan, Azerbaijan, Iran, Afghanistan, Syria, Iraq, Greece, Kyrgyzstan, Kazakhstan and Kosovo rank among the top ten countries which send the most number of students to Turkey in 2013-2014 (Kadıoğlu & Özer, 2015).The students from these countries first learn Turkish at the Turkish Language Teaching Centers at the universities and then continue to their education.Additionally, because of the political and economic developments, it has been observed that recently Turkish Education has increased in remote regions as well.
The increase in the teaching of Turkish as a foreign and second language has brought diversity in course materials, teacher education programs and academic studies.Academic studies, which previously identified Turkish learners as a homogenous population, have begun to become more and more specialized in a way as to take into consideration learners' countries, mother tongues and specific needs.Master and doctoral theses related to the Turkish education of the individuals whose mother tongue is Bosnian, Arabic, Persian, and Russian and so on are being conducted and special course materials are being developed.
In this context, the concept of linguistic distance is defined as a feature that should be taken into account when studies and developed materials are concerned.For example, the features of learning in the case of a person who speaks Mongolian as the native language with deep historical ties with Turkish and in the case a person whose mother tongue is Tagalog in the Philippines, where the Turkish language has just arrived are different.In this context, it is necessary to reveal through academic research how effective the mother tongue variable is in Turkish education.Despite the studies related to this issue (Bölükbaş, 2011;Er, Bicer & Bozkırlı, 2012;Karababa, 2009;Köse, 2015;Subaşı, 2010), the topic has not been examined in the context of language families and linguistic distance.Accordingly, the aim of this research is to examine the scores of the international students registered in the Turkish Language Teaching Centres located at the Gazi and Hacettepe Universities in terms of age, gender, and language family and region variables.
In order to reach the purpose of the research, the following questions have been answered:

Research Design
This study, in which the total course final scores of the Turkish language learners are studied in terms of the mother tongue variable, was carried out using the relational screening model in accordance with the quantitative research."The relational screening model is a research model aiming the determination of the presence or degree of variance between two or more variables" (Karasar, 2009, p.81).

Data Collection and Analysis
The data for this study were obtained from Turkish language course completion exams and personal information forms.The course completion exam is the final exam at the end of each academic year at the Turkish Language Teaching Centres at Gazi and Hacettepe Universities.Registered students continue their education at university by getting a certification according to the score they get at the end of the exam.These exams consist of speaking, writing, reading comprehension and grammar parts.The average score of the students taken from these sections is considered as their grade.The data collected for the research were gathered from the exams conducted in 2015 and 2016.Besides, personal information forms included information on age, gender, country, mother tongue, enrolment type and university information of the students.
Direct frequency analysis was utilized for the age, gender, and enrolment type and university information obtained from personal information forms.The countries of the students were grouped in terms of their geographical, cultural and linguistic characteristics, and their native languages were grouped in terms of their language families and then subjected to frequency analysis.
Multiple regression analysis was conducted on the variables such as age, gender, language family and region along with the speaking, writing, reading and grammar knowledge and average grades got in Turkish language course final examinations by the international Turkish learners.In regression analysis, the dependent and independent variables must be continuous variables measured in the same interval scale and should be in normal distribution.However, some might aim to exceptionally examine the effects of the independent variables on the dependent variable included in the classification scale.A classification variable in the analysis is created by forming a new artificial variable called a dummy variable through excluding one of its levels produced by a subtraction of the number of levels (G-1).That one of these new variables has a significant impact on the dependent variable can be interpreted as an important effect of the independent variable on the dependent variable (Büyüköztürk, 2012, p.92).
In that sense, the discrete variables in the study were included in the regression analysis as a dummy variable.The categorical variable was handled in eleven categories and the Balkans category was coded as "0" to make a dummy variable.The variable for the language families is grouped into nine categories and the Altaic category is coded as "0", and dummy variable was formed.In the gender variable, the female category was coded as "0" and made the dummy variable.Since Turkish language is belonged to the Altaic language family, this language family was chosen as a reference.
There is no specific reason for the selection of the Balkans.Considering the gender variable, it is taken into account that women might be more successful in Turkish language exams than men.Apart from this, nothing has been carried out for discrete variables.After organizing the data, predictive ability of the variables related to the scores of the students from the course completion exam was tried to be determined through the multiple regression analysis.

Findings
The results of the multiple regression analysis that was done by using the collected data in the scope of the research questions are presented below by dividing them into the categories such as mean scores and skills of writing, speaking, reading and grammar.

Findings Related to Average Scores
The results of the multiple regression analysis that was conducted to determine how the variables such as age, gender, language family and region predict the mean score of the Turkish language course exams of the international students are presented in Table 4.As a result of the multiple regression analysis age, region (Eastern Europe) and, Austronesian, the Afro-Asiatic, Bantu, Sino-Tibetan, Indo-European language variables were found to significantly predict the mean score of the students.
It is clear that the correlation is R=.43 and variance is R 2 =.19 for this regression model (F (13−266) =4.72, p<.01).The mentioned variables explain approximately 19.00% of the variation in the average test scores.According to the standardized regression coefficients, Afro-Asiatic (β=-.47)language family appears to be the first when the relative importance of the predictive variables over the mean scores is considered.When the t-test results related to the significance of the regression coefficients are considered, it is obvious that Austronesian, the Afro-Asiatic, Bantu, Sino-Tibet, Indo-European language families, Eastern Europe region and the mean scores of age are statistically significant predictors among all predictive variables.
Similarly, when the language family, age and gender variables are handled, the scores of students from Eastern Europe (B = 8.49, β=.14) are higher than those students coming from the Balkans.Lastly, it was seen that as the age of the students increased, their exam scores also increased in a statistically significant way.

Findings Related to Writing Scores
The results of the multiple regression analysis for the writing scores of international students in the Turkish course completion test are presented in Table 5.The results in Table 5 show that with the age, Austronesian and the Afro-Asiatic language families were found to statistically predict the students' writing scores.It is seen that this regression model is R=.44, R 2 =.19, (F (13−266) =4.96, p<.01).These variables explain almost 20.00% of the variation in writing scores.
According to the standardized regression coefficients, when the relative importance of the predictive variables over the writing scores is considered, the Afro-Asiatic (β=-.44)language family appears to be the first, and it is followed by Austronesian (β=-.21)language family and age (β=.21).When the t-test results on the significance of the regression coefficients are handled, it is apparent that age as a predictive variable is a significant predictor over the writing scores of Austronesian and the Afro-Asiatic language families.

Findings Related to Speaking Scores
The multiple regression analysis results for speaking scores from the Turkish language course exams of the international students are presented in Table 6.According to the results presented in Table 6, it is seen that along with the age variable, the Austronesian, Afro-Asiatic, Bantu, Sino-Tibetan, Indo-European language families and Eastern Europe's region variable's regression coefficients statistically significantly predicted speaking scores of the students.It is seen that this regression model is R =.45, R 2 =.20 (F (13-266) =5.23, p<.01).These variables clarify 20.00% of the speaking scores.According to the standardized regression coefficients, when the relative importance of predictive variables on speaking scores is examined, Afro-Asiatic (β=-5.03)language family gets the first.Indo-European (β=-.40) and Bantu (β=-.33)language families follow that.
When the t-test results for the significance of the regression coefficients are examined, it is seen that the predictive variables of Austronesian, Afro-Asiatic, Bantu, Sino-Tibetan, Indo-European language families are statistically significant predictors of Eastern Europe region and age speaking scores.

Findings on Reading Comprehension Scores
The results of the multiple regression analysis to determine the variables that predict the reading comprehension scores of the international students in the Turkish language course completion exams are given in Table 7.According to these results, it was seen that age and region variables of the Afro-Asiatic, Bantu, Sino-Tibetan and Indo-European languages families were statistically significant predictors of the students' reading comprehension scores.This regression model was found to be R=.38,R 2 =.14(F (13−266) =3.38, p<.01).These variables clarify 14.00% of the grammar scores.According to the standardized regression coefficients, Afro-Asiatic language family (β=-.27)appears to be first when the relative importance of the predictive variables over the reading comprehension scores is examined.This is followed by Indo-European (β=-.25) and Bantu (β=-.23)language families.When the t-test results for the significance of the regression coefficients are considered, it is seen that the predictive variables, such as age and Austronesian, Afro-Asiatic, Bantu, Sino-Tibetan and Indo-European language families are statistically significant predictors of reading comprehension scores.

Findings Related to Grammar Scores
The results of the multiple regression analysis to determine the variables that predict the grammar scores of the students in the study group are given in Table 8.The results of the multiple regression analysis show that Austronesian, Sino-Tibetan language families, Eastern Europe region variable statistically significantly predicted the grammar scores.It was found out that this regression model is R=.32, R 2 =.10(F (13−266) =2.38, p<.01).These variables clarify 10.00% of the grammar scores.
According to the standardized regression coefficients, when the relative importance of the predictive variables on grammar scores is examined.It is seen that the Sino-Tibetan (β=-.17)language family appears to be first.This is followed by Austronesian (β=-.16)language family and Eastern Europe as region variable (β=.14).When the t-test results on the significance of the regression coefficients are considered, it is seen that there is a significant predictor on the Austronesian and Sino-Tibetan language families and Eastern Europe region's grammar scores.

Discussion, Conclusion & Implications
This study, which is based on the belief that mother tongue and its language family are effective in the language learning of the individual, has investigated to what extent the mother tongue of the international students who come to Turkey to learn Turkish has influenced their success.Within this scope, the students' mother tongues were grouped according to their language families and multiple regression analyses were carried out in order to show how effective the language families are on the scores students get from the Turkish language course completion exam.Even though the results of this study cannot be generalized universally, they give an idea of how the Turkish language learners are influenced by the mother tongue variable.
In this study, all findings related to the language family were interpreted by basing onto the Altaic language family.Accordingly, the mean scores that students get from the Turkish language course completion exam are lower for the Afro-Asiatic, Indo-European, Bantu, Sino-Tibetan and Austronesian language families than Altaic language family.
In terms of language skills and grammar scores of participants, the conclusions of the study: The writing scores of the participants whose native language belongs to the Afro-Asiatic or Austronesian language families are statistically lower than the ones which belong to the Altaic language family.As for the speaking scores, they are lower for Austronesian and Sino-Tibetan languages, and mostly by Afro-Asiatic, Indo-European and Bantu language families compared to Altaic language family.Also, it has been found out that the reading comprehension scores of the participants whose mother tongue belongs to the Afro-Asiatic, Indo-European, Bantu and Sino-Tibetan language families are negatively affected more when compared to the students' scores whose native language belongs to the Altaic language family.The grammar scores are lower for students coming from Sino-Tibetan and Austronesian language families than the Altaic language family.
If the findings are interpreted in the context of the language families, the findings are similar to the literature review which indicates that when compared to other language families, the Afro-Asiatic language family has more influence on participants' writing, speaking and reading comprehension scores they receive from the exams than the Altaic language family.Bölükbaş (2011), Er, Biçer and Bozkırlı (2012), Karababa (2009), Subaşı (2010) remark that Arabic students have a great deal of problems in writing because of the difference in alphabets and in speaking due to the sounds that don't exist in their mother tongue.In addition to this, Kara (2010) points out that Arabic students have more problems in reading and writing Turkish than other international students have.Findings are consistent with the literature review; that is, speaking, reading comprehension, and grammar scores of Sino-Tibetan language speakers, such as Chinese, are influenced negatively more than Altaic language family speakers.Köse (2015) who states that Chinese students have difficulty in distinguishing Turkish syllabus concluded that these students are mistaken in Turkish word stress and cannot pronounce some sounds.Also, teachers have encountered some problems especially in the teaching of Turkish compound tenses because the grammatical features of Turkish and Chinese are different.
The fact that speaking and reading scores of the Indo-European language family participants are affected negatively more than the ones in Altaic language family is similar to the findings gathered by Tüm (2014) in the research on speaking skills of European students speaking a language belonged to this family.The students have problems in Turkish sounds, which do not exist in their mother tongue.
While it was observed that African students were successful in the grammar lessons, Kara (2010) indicates that this achievement was not seen in speaking and writing lessons and also it was found out that speaking and writing scores of the students of Bantu language family are affected negatively more than the participants whose mother tongue belongs to the Altaic language family.However, the lack of academic studies on the problems encountered in the teaching of Turkish to the native speakers of Bantu languages makes it difficult to interpret the findings.Similarly, it is necessary to conduct more academic research on the languages of the Austronesian languages, such as Malay and Indonesian which have lots of speakers, and writing, speaking.Additionally, grammar scores of Turkish by the students speaking those languages as native languages are negatively affected more when compared to the Altaic language family.
Apart from the mother tongue, the age variable influences the mean score, writing, speaking and reading scores except grammar scores.Participants' ages positively affect the writing, reading and mean scores they got from the Turkish course completion exam, and on the other hand, it affects the speaking scores negatively more than it affects the ones in the Altaic language family.In other words older participants tend to have higher scores for writing, reading and mean scores, but younger participants' tend to have higher speaking scores.This suggests that students starting language learning at very early ages are more successful in speaking than the learners who start it late.Besides, gender and region variable do not predict the success.
That the variance of the scores are not too high shows that language families are not the only variable to explain the success of the students in Turkish language course completion exams, but it is a factor affecting the success.This finding is too similar to the results of the research by Elder and Davies (1998) that English learners' success in the certificate examinations cannot be explained only by the mother tongue.
It was concluded in the study that there is a significant difference between the participants' writing, reading, reading comprehension, grammar and mean scores from the Turkish language course completion exam and the language families.In this context, the findings of the study are consistent with the assumption of Corder (1978) that the case in which there are many similarities between the mother tongue and the foreign language being learnt is more advantageous than the case where there is little similarity.The findings of the study, which are thought to be preliminary studies, are needed to be examined deeply with the future qualitative studies.

Table 1 .
Demographic Characteristics of the Participants.

Table 2 .
Distribution of the Participants by Region.

Table 3 .
Distribution of Participants' Native Languages by Their Language Families.

Table 4 .
Results of Regression Analysis of the Factors Predicting Mean Scores.

Table 5 .
The Results of Regression Analysis of the Factors Predicting Writing Scores.

Table 6 .
The Results of Regression Analysis of the Factors Predict Speaking Scores.

Table 7 .
Results of Regression Analysis of the Factors Predicting Reading Comprehension Scores.

Table 8 .
The Results of Regression Analysis of the Factors Predicting Grammar Scores.