Can Late Starters of L2 Achieve Native-like Pronunciation?

This study examines the whys and wherefores of nonnative speech of Turkish which may result as heavily accented or native-like accented for the late starters of Turkish. Late start in this context is used to notify that none of the learners started learning Turkish during puberty. The experiment was held at the University of Texas. The subjects were 5 students who are currently enrolled in Turkish classes at U.T. Three of them are enrolled in 2nd year Turkish class, one is enrolled in 1st year Turkish class and the remaining one is a 3rd year Turkish student whose ages ranged from 21 to 67. Three of them speak English and one of them speaks German as his/her native language. I asked four native speakers of Turkish to listen to the recorded words which were formerly produced by the 5 subjects about whom I mentioned above. The judges were asked to rate them for the degree of foreign-accentedness and comprehensibility on 9 point scales. There seems to be no correlation between the accentedness and the comprehensibility when one looks at the results. All of them claimed that even the most heavily accented subject could be understood without any difficulty. None of the judges had any experience in linguistics. Pronunciation and my experience As a nonnative speaker of English, I spent most of my time on listening to music with English lyrics and watching authentic materials of the foreign language while I was learning English at school, which is English in my case. I started learning English at the age of 13 which is accepted as the end of critical period by most of the researchers. From my point of view, becoming saturated with the foreign language via listening to and watching materials improved my pronunciation in many ways. As a foreign language instructor in my country, I tried to use visual and auditory material during my classes as much as possible. I had my learning of how to pronounce as a model. Mimicking helps to have a native-like accent and to be able to mimic, being saturated with authentic auditory and visual material holds an important role. To learn if the subjects spent any time on listening to music with Turkish lyrics and watching things in Turkish, I conducted a survey which includes questions about their language background, time and its frequency spent on authentic materials, length of stay in Turkey, frequency and length of Turkish use outside the class, whether they had any Turkish parents, relatives, partners or friends, length of Turkish education they had so far and since when they’ve been familiar with the language both in formal and informal settings. The results of this survey helped me to correlate the scores which were given by the inexperienced native speakers. Method Speakers and Survey 5 Nonnative Speakers of Turkish (3 Female and 2 male) participated in the recording. All of whom started learning the target language at U.T. None of them has or had any fascination towards the target language or culture. I’d like to give some background information depending on their answers to the survey which they filled out and e-mailed me back. Their Turkish education ranges from 3 semesters to 3 years. The duration of their daily Turkish practice outside the class ranges from 5 to 20 minutes. The frequency of listening to music with Turkish lyrics ranges from once a day to once a week. Three of the subjects have been to Turkey and their duration of stay ranges from 3 months to 3 years. Their becoming familiar with the target language dates back to 2 years for three of them, 3 and 4 years for the other two. One of the subjects is from Germany and grew up in Germany. She had never tried to learn it in Germany although had had many Turks around her. The rest have been familiar with Turkish since they started taking classes at the University of Texas. When asked if they have any Turkish parents, relatives, partners or friends. Three of them reported they had friends, one of them replied no and one reported that he his fiancé is Turkish.Their weekly exposure of Turkish (In class, watching or listening to things in Turkish, chatting with a native speaker, etc.) ranges from 4 hours to 8 hours. All the subjects speak languages other than Turkish and their native languages. Two of them speaks German and evaluate themselves as having good pronunciation. Another subject speaks English, French, and Spanish and studied Latin additional to her native language German. She evaluated her pronunciation of English as ‘foreign accented’. The other two have Spanish in common whereas one of them speaks Italian additionally. The one who speaks both Spanish and Italian claimed to have very little knowledge of the languages. The subject who speaks Spanish only evaluates his pronunciation as ‘not as good as it used to be’. Another question in the survey was if they found it hard to pronounce the words in Turkish or not and how they could improve their pronunciation. Two of the participants claimed that they sometimes find the words hard to pronounce. One of them said that improving one’s accent could be through hearing a native speaker and practicing with one. The other subject who sometimes found it hard to pronounce claimed that reading aloud is the way to improve it. One of the remaining ones said that it wasn’t hard and the way to improve it could only be by listening to native speakers, speaking to native speakers. The last participant reported that pronouncing words in Turkish was hard and he could improve it by practicing. Recording Individual recording sessions were held in a sound treated room. A tape recorder was used. The participants were to read aloud the words which were shown on the computer screen during the first session and they were to repeat another list of words after hearing my pronunciation during the second session. They were allowed to look at the words during the second session as well. The entire task took two minutes approximately for each subject. Listeners The listeners were 4 native speakers of Turkish who live in Austin. One of them is an Architect, another one is a Research Assistant at the Department of Computer Engineering. The remaining two are students who study at the Faculty of Engineering at the University of Texas.They neither studied nor have a background of linguistics or language teaching. Procedure I held two listening sessions one after the other. I had 2 of the judges evaluate the recordings individually and the other two together in quiet rooms around the campus. The listeners were to rate the accent and the comprehensibility of the subjects’ uttering 6 Turkish words in the first session. I stopped the recording after each word which was pronounced by the subjects. They were free to listen to the recording for the second time during both of the sessions. The listeners were to grade the subjects on a 9-point scale, where 1=no foreign accent or extremely easy to understand and 9= very strong foreign accent or impossible to understand for the accent and the comprehensibility evaluation. The first session lasted approximately 4 minutes. In the second session the judges listened and evaluated the subjects’ accent and comprehensibility as they heard them mimicking 7 words which were different from the first set of words pronounced in the first session. The second session took approximately 5 minutes. Results I calculated the mean of accent and comprehensibility scores of each subject which were given by the listeners. The uttering scores, which the participants were graded based on their pronunciation without any help ranged from 3.5 to 5.0. The comprehensibility scores of the same set of words ranged from 2.0 to 4.0. It is not surprising to see that scores show their having less foreign accent in the comprehensibility task, as all the listeners claimed that no matter how foreign accented the subject was they were able to understand without any difficulty. The mean of the accent scores of the subjects’ mimicking the words ranged from 2.5 to 3.5. As expected the mimicking scores revealed the subjects’ being less foreign accented than the mean of the uttering scores in the first session. The mean of the comprehensibility scores of the mimicking session ranged from 1.50 to 3.50, which were again, not surprising according to what the judges claimed about the comprehensibility task. The participant, who was reported as the least foreign accented (3.5) subject during the first session, is the one who speaks 5 languages including Turkish. She is 67 years old and had no time to spend on target language material outside the class. She grew up hearing Turkish in her home country, Germany. She lived in Turkey for 3 years. I assume that the reason why she was evaluated as the most native-like accented subject could be that, among the other subjects, she spent the longest time in the target country. She was also evaluated to be the best scorer in the comprehensibility task both during uttering in the first session and mimicking in the second session. This proves that there’s a positive correlation between the accent scores and the comprehensibility scores. The participant, who was reported to be having the best score (2.50) at mimicking, is 21 years old and lived in Turkey for 4 months. She studied at a Turkish university during that time. According to the survey which the participants filled out, among the other subjects, she spends more time on daily practice than the other participants do. She was the one who didn’t find it hard to pronounce Turkish words. Her reply to the question of how to improve her pronunciation was listening and speaking to the native speakers. On the other hand, her score of comprehension, in terms of both uttering in the first session and mimicking in the second session, was the second most native-like one. Discussion The results of this study seem to be close to what I expected. My assumption was that at least one of the participants could be evaluated with “no foreign accent” during the mimicking session. However none of the participants could score a native-like accent or be categorized as extremely easy to understand. Based on the fact that all of the participants started learning Turkish after the critical period, this study could at least give an overall idea that it is not impossible to achieve a close score to the level of a native-like accent. Furthermore one could tell that mimicking helps a learner to attain better pronunciation as the subjects were evaluated to have a better native-like accent in the second session when they are compared with their uttering scores in the first session. The difference between the uttering scores and mimicking scores ranged from 0.75 to 2.25. The score difference which reaches 2.25 reveals that there’s a slight difference between the subject’s word utterance without any help and her repeating another word after the native speaker. The same subject scored 4.75 when she uttered the first set of words in the first session and scored 2.50 when she mimicked another set of words.