Singing Talent and Foreign Language Pronunciation: An
Empirical Study on Native Chinese Speakers
Name: Decai Xia
Student number: S3104257
Supervisor: Dr. Charlotte Gooskens
Second reader: Dr. Dicky Gilbers
Date of completion: June 4th, 2017
Lists of Tables and Figures...IV Acknowledgment... V Abstract...VI Chapter 1. Introduction... 1 1.1 Introduction... 1 1.2 Thesis Overview...2
Chapter 2. Theoretical Background... 3
2.1 Definitions...3
2.1.1 Singing Talent...3
2.1.2 Musical Aptitude and Musicality... 4
2.1.3 Working Memory and Phonological Loop...5
2.2 Music and Language... 6
2.3 Singing and Speaking...8
2.4 Studies Confirming Musical Aptitude/Singing-Foreign Language Pronunciation Relationship... 9
2.5 Studies Denying Music Aptitude-Foreign Language Pronunciation Relationship...13
2.6 Summary of Studies... 15
2.7 The Present Study...16
Chapter 3. Research Methodology...17
3.1 Participants... 17
3.2 Instruments... 18
3.2.1 Questionnaire...18
3.2.2 English Reading Task and Sentence Imitation Tasks... 18
3.2.3 Singing Talent Test... 20
3.3 Procedures... 21
3.3.1 Questionnaire Survey... 22
3.3.2 English Reading and Sentence Imitation Tasks...22
3.3.3 Singing Talent Test... 23
3.3.4 Working Memory Test...23
3.4 Data Preparation and Analysis... 24
Chapter 4. Results... 25
4.1 Correlation between Singing and Foreign language Pronunciation...25
4.2 Correlation between Language Pronunciation and Working Memory...27
4.4 Subjects Divided by Gender...29
4.5 Other Variables in the Questionnaire...30
Chapter 5. Discussion...33
5.1 Singing and Foreign Language Pronunciation... 33
5.2 Working Memory and Foreign Language Pronunciation...36
5.3 Gender Difference in Foreign Language Pronunciation... 37
5.4 Age of Acquisition and Frequency of Use...38
Chapter 6. Conclusion...39
6.1 Major Findings... 40
6.2 Implications...40
6.3 Limitations and Directions for Further Research...41
References... 42
Lists of Tables and Figures
Figure 1. Flowchart summarizing the entire experiment process... 22
Figure 2. Correlation between English reading and singing... 27
Figure 3. Correlation between English imitation and singing...27
Figure 4. Correlation between Catalan imitation and singing...27
Figure 5. Correlation between English reading and working memory... 28
Figure 6. Correlation between English imitation and working memory...29
Figure 7. Correlation between Catalan imitation and working memory...29
Figure 8. Means of females and males in four language production tasks... 30
Figure 9. Means of females and males in the seven variables in the questionnaire..31
Table 1. Correlation between singing and pronunciation...25
Table 1. Correlation between singing and pronunciation...25
Table 2. Correlation between working memory and language pronunciation... 28
Table 3. Correlation between each variable and language pronunciation...32
Table 4. Multiple regression indicating the strongest indicators for pronunciation in English imitation... 33
Acknowledgment
Abstract
Chapter 1. Introduction
1.1 Introduction
Due to the increasing demand for global communication, the acquisition of a second language, especially English as the global lingua franca, has become more urgent than ever. However, second language acquisition including the mastery of vocabulary, grammar and phonology seems to be much more difficult compared to the natural acquisition of first language which starts even before birth. To pronounce accurately is one of the most difficult skills in learning a foreign language and is of vital importance, as it can predict the proficiency of speakers in this language and ensure a comprehensible communication (Dalton, Kaltenboeck and Smit, 1997). Individuals differ from each other in their ability to pronounce accurately even though other factors related to second language learning such as age of acquisition, quantity and quality of language exposure and language use are similar, resulting in an assumption that there is an innate or acquired skill or ability that can contribute to people’s acquisition of an accurate and native-like pronunciation of foreign languages (Sleve and Miyake, 2006; Rota and Reiterer, 2009).
Further, I have observed such a phenomenon in my past experience in learning foreign languages that teachers who sounded native-like in their spoken language could also sing well and the same goes for my peer classmates. On the contrary, those peer classmates who struggled with but made little progress in their pronunciation mostly considered themselves as having no talent in music and singing at all. This thus triggered me to wonder whether there is some connection between the ability to sing and phonetic acquisition of a second language. A brief review of literature related to this topic, which will be discussed in details in chapter 2, encouraged me to explore further. Recent empirical studies have proved that musical talent is a predictor of a good pronunciation of foreign languages, and there is also an increasing number of investigations, showing that the verbal information and music may be processed in similar locations in the brain (Koelsch et al., 2009; Williamson, Baddeley & Hitch, 2010; Schulze and Koelsch, 2012). Since singing is frequently considered as a subcategory of music, this study will focus on the relationship between singing talent and pronunciation of foreign languages.
standardized tests based on the view that musical aptitude consists of a number of separate abilities such as pitch discrimination, rhythm sensation, sense of harmony and tonal memory (Gilleece, 2006; Dolman and Spring, 2014). Therefore, they are used to measure the participants’ ability to discriminate changes in rhythm, tone and other musical variables. Such studies focus purely on the perception of music rather than production of what is perceived. Those including the study of musical production only require for the reproduction of short simple melodies or musical notes, which can be easily done by most people and cannot examine people’s ability to reproduce melodies with complex note and pitch combinations that require for greater control over vocal motor system (Christiner and Reiterer, 2013). As a matter of fact, human naturally develops the ability to sing in earlier years of development, stabilizing at around the age of five if receiving no special training and thus, ordinary people even without training can sing familiar songs or repeat short new melodies, though not always precisely (Dalla Bella et al., 2012). Notwithstanding the study of Christiner (2013) combined both AMMA test and singing performance, they focused on musically-trained people and studies on the influence of musical production in the form of singing on musical naive people can hardly be found.
Considering the paucity of existing information on the what the role of singing talent plays in the phonetic acquisition of foreign language, it is necessary to explore their association further. This study aims to investigate whether the positive relationship between singing ability and foreign language pronunciation found on vocalists in the research of Christiner (2013) is also valid on ordinary people without special training on singing, together with an investigation on some extra-linguistic variables such as age of acquisition and frequency of use. It is also hoped that this study can make a contribution to the growing knowledge of the relationship between singing and second language pronunciation. Further, a good working memory is proved to be crucial for the success in phonetic acquisition of foreign language. Hence, the role which working memory plays in the phonetic acquisition of foreign languages will also be investigated. A deeper understanding of the correlation between singing, working memory and foreign language pronunciation can help language learners formulate a more effective learning strategy and inform language teachers the importance of these abilities in improving pronunciation of target languages.
1.2 Thesis Overview
singing performance and foreign language learning. The research questions and hypotheses are also included in this chapter. Chapter 3 describes the research methodology used in this study, including participants, materials, procedures and data preparation and processing. Chapter 4 shows the results of this study. Chapter 5 discusses the results based on the literature reviewed in Chapter 2. Chapter 6 summarizes the main findings and conclusions of this study and points out the limitations of this study and directions for further research.
Chapter 2. Theoretical Background
In this chapter, some definitions related to this study such as singing talent, musical aptitude, musicality, working memory, relationship between music and language and singing and speaking will be firstly discussed, and then literature concerning the relationship between music aptitude/singing and foreign language pronunciation, that between working memory and foreign language pronunciation, gender difference in foreign language pronunciation, the effect of age of acquisition and frequency of use on pronunciation will be presented.
2.1 Definitions 2.1.1 Singing Talent
When talking about singing, what cannot be ignored is music, as the former is often considered a subcategory of the latter, though most of the researchers did not make a clear distinction between them and blurred their differences. More often than not, except those created purely by musical instruments, a significant number of music pieces is achieved by both musical instruments and singing, and thus singing is one way of musical production highlighting the use of tonality, rhythm and some other vocal techniques. This, to some extent, explains why singing is a subcategory of music but not vice versa. Previous studies on speech imitation have proved that the ability to sing well is a much stronger indicator of accurate pronunciation and native-like accent in second language acquisition than the ability to play an instrument (Christiner and Reiterer, 2013).
general intelligence or mental power (ability)” (p. 214). Further, talent is considered as an “exceptional capability” in a specific field as well as a potential to develop further (Jørgensen, 2008, cited in Nardo and Reiterer, 2009, p. 213). As a result, developing out of the definition of musical talent that it is “a (predominantly) innate tendency to understand/appreciate, perform or create music outstandingly” (Nardo and Reiterer, 2009, p. 214), singing talent is an innate ability to perform and sing well or make some achievements in singing. Talent in singing is not evenly distributed among the general population and cannot be developed out of experience or training, but it does not mean that people not talented in singing do not possess the ability to sing, as it can be acquired by training and practicing. Since the singing ability of people receiving no training is purely innate, singing ability is considered as the synonym of singing talent in this study.
2.1.2 Musical Aptitude and Musicality
In addition to musical aptitude, some other terms like musical talent as mentioned above and musical intelligence bearing similar concepts with musical aptitude as well as musical ability and musicality referring more to ability can be found in use in exiting literature.
As defined by Carroll (1990), general aptitude is the “capacity of learning a task, which depends on some combination of more or less enduring characteristics of the learner” (p. 84). Music aptitude, according to Gordon (1987), is “a measure of one’s potential for music learning and foundation of the music achievement” (cited in Pei, 2010, p. 6). It is an ability people are born with though levels of it varies from one to another. A person’s musical aptitude keeps on developing until the age of nine and stabilizes after that (Gordon, 1987). What needs to be pointed out is that the music achievement of a person does not always reflect his or her musical aptitude. It is true that those with high levels of music achievement must accordingly have a high level of music aptitude, whereas the opposite is not necessarily true, as a person with high music aptitude may have a lower music achievement than their aptitude allows them to, if they are never encouraged to participate in music (Gordon, 1987). Thus people with high musical aptitude, given proper instruction and training, can have better music achievement even after critical age than those with low level of musical aptitude who although have already had a high music achievement (Gordon, 1988).
The term musicality is widely used in describing and encompassing all aspects of the field of music (Christiner and Reiterer, 2013). Most researchers believe that musicality is innate and can be inherited, and similarly, it is considered as similar to musical ability, both referring to a partially innate skill that can be strengthened by interacting with the environment such as training and practicing,, with the emphasis
form of rhythm perception, pitch perception, ability and liking for singing (Nardo and Reiterer, 2009). Thus singing is often considered as a standard to measure musicality is art education.
2.1.3 Working Memory and Phonological Loop
A wide range of studies (Rota and Reiterer, 2009; Baddeley, 2003; Pastuszek-Lipinska, 2008) has proved that working memory, especially phonological working memory, is a factor responsible for whether a person can have near-native attainment in second language acquisition and it is also “an important component of the cognitive processes underlying bilingual language processing and performance on measures of L2 proficiency” (Linck et al, 2014, p. 879).
However, before discussing working memory, it is necessary to have an overview of what memory is. According to Whingfield and Byrnes (1981), the term memory “has to do with the capacities human beings have to retain information, to recall it when needed, and to recognize its familiarity when the later see or hear it again” (p.4). Memory can generally be divided into sensory memory, short-term memory and long-term memory. Sensory memory, as the shortest-term memory, serves as the buffer for information newly received through sight, hearing, smell, taste and touch (Whingfield and Byrnes, 1981). Short-term memory, with limited capacity, can only store information for a short period. This capacity has been determined and measured in many ways, of which digit span, as a subtest of Wechsler Adult Intelligence Scale (Wechsler, 1997), and non-word repetition, is commonly used (Kormos and Sáfá, 2008). It is believed that information stored in short-term memory has been coded into some form of language. On the contrary, long-term memory is not constrained by capacity and seemingly can store unlimited amount of information for indefinite time. It involves more of the semantic aspect of the information.
the fourth subsystem, the episodic buffer dedicated to integrating all the information with chronological ordering into this model.
In the model of Baddeley and Hitch, the phonological loop, consisting of a phonological store, which serves as an articulatory rehearsal process, and articulatory rehearsal process, is the structure responsible for language acquisition (Baddeley, 2003; Atkins and Baddeley, 1998). While the phonological store can retain information for seconds, the articulatory rehearsal process resembling the sub-vocal speech allows for a longer maintenance through active rehearses such as repetition of verbal information. The basis for human’s ability to spontaneously repeat new phonological strings and then to memorize them is formed by the phonological loop (Rota and Reiterer, 2009). Further, the phonological loop can store “the representation of new linguistic material and structures” temporarily (Rota and Reiterer, 2009, p.17), and the capacity of phonological loop can predict the ability of L2 acquisition after the critical periods (Atkins and Baddeley, 1998), which is evidenced by a patient suffering from a deficit of phonological loop, who failed to learn vocabularies of a new language.
To prove the relationship between phonological short-term memory and foreign language performance and working memory and foreign language performance, Kormos and Sáfá (2008) invited 121 secondary school students from Hungary to participate in a non-word repetition task, measuring phonological short-term memory and Cambridge First Certificate Exam testing both reading, writing and speaking abilities, and 50 of them also took a backward digit span test, which measured their working memory. A significant correlation was found between backward digit span and speaking performance, r = 0.33. Such a correlation between working memory and speaking, as explained by the authors, is due to the fact that “speaking in L2 is the task that poses very high attentional demands on learners” (Kormos, 2006, cited in Kormos and Sáfá, 2008, p. 268), whereas “working memory is also responsible for regulating attention in cognitive processing” (Baddeley, 2003, cited in Kormos and Sáfá, 2008, p. 268).
2.2 Music and Language
cited in Morgan, 2003). Similarly, Darwin (1871) claimed that “musical sounds afforded one of the bases for the development of language” (p.639, cited in Dolman and Spring, 2014) and music and language also have the same origin. Besides the hypothesis of the musical origin of language and same origin of language and music upheld by Rousseau (1781, cited in Morgan, 2003), Darwin (1871), and more recent scholars Vaneechoutte and Skoyles (1998) who proposed the musical origin of language, other opinions about where language and music originate from can also be found in existence. For example, Levman (1992) mentioned two other major propositions in his work: 1) language and music are two different faculties and develop separately; 2) music developed out of language or music appeared later than language.
Despite those divergent opinions on the origin of music and language, many researchers have reached an agreement on the commonalities between language and music from a cognitive perspective. For instance, both music and language are an auditory phenomena that perceive and produce sound patterns by auditory and oral organs, and they are functioned to express the emotions and convey messages by the variations in the intonation of voice. (Christiner, 2013; Trehub, 2000, 2001, cited in Gilleece, 2006). Both of them are organized temporally and consist of discrete elements structured hierarchically (Nardo and Reiterer, 2009; Jackendoff, 2009). Both of them are alike in many intrinsic aspects like pitch, stress, tone, rhythm and pauses (Mora, 2000), even though the rhythm in music and language is different. More importantly, musical abilities appear to explain partially how language is acquired and processed, reflected in infants’ language acquisition. They recognize when the words, phrases or sentences start and end, and probably learn the meanings of words through the prosodies of language, such as melody, rhythm and timber of the speech (Jusczyk, 1999).
In addition to structural and functional similarities, neuroscientific studies have also proved that the processing of music and language happens, at least partially, in the same brain structures. Although the very earliest such studies (Milner, 1962; Kimura, 1964) suggested that language was processed in the left hemisphere of the brain, while music processing was done in the right, later evidence indicated that the prosodic features of language like intonation and accentuation are processed in the right side of the brain, and the syntactic and semantic aspects of language are processed in the left (Dehaene et al, 1999). Moreover, the study of Patel et al (1998) on musicians, which compared ERPs elicited by incongruent syntactical structure in both music and language showed that the same P600 elicited by linguistic incongruities specific to language, could also be elicited by musical incongruities.
Even though music and language share a series of similarities, it does not mean that there is always a direct equivalence in both, as similarities can also be interpreted as differences on a different ground of analysis. For example, the metric structure in musical pieces is consistent and specific, but the prosodic structure is changeable in language (Besson and Schön, 2001). Whilst the musical system is made up of pitches and timbers, the speech sounds of the linguistic system consists of “vowels, consonants, and pitch contrasts of the native language” (Patel, 2008, p.9). More distinctions can also be found in syntactic and semantic levels, in which the utterances in language are formed by words and syntax, but on the contrary, music pieces are built up with individual tones, formulaic patterns and prolongational structure (Jackenoff, 2009).
2.3 Singing and Speaking
Unlike musical aptitude, musicality and fields as such which have been investigated extensively, the singing talent of ordinary people has not been studied much, despite of the fact that singing is one of the most widespread ways to express music. Though it is generally believed that singing abilities are not equally distributed among ordinary people, later studies, however, proved that nonmusicians showed a highly consistent ability in singing familiar songs and people tend to underestimate their ability to carry a tune (Peretz, Gagnon, Hébert & Macoir, 2004). While people generally consider singing as a subcategory of music, some studies (e.g. Christiner and Reiterer, 2013) conclude that it is closer to speaking and can act as the “parasite of speaking” (Rebuschat et al, 2011, p.5).
and speaking share some basic elements, for instance posture, breathing, resonance and articulation (cited in Christiner and Reiterer, 2015). What makes the singing and speaking process different is that more active exhalation than inhalation is needed in singing than speaking, and the “vocal technique pursues correct glottic closure with a stable larynx position, which is achieved by the coordination of not only the extrinsic but also the intrinsic muscles” in singing (p. 443, cited in Christiner, 2013, p. 46).
Since a song is composed of melody and text, it can be inferred that singing results from both music processing and language processing, and “in song, music and speech are integrated rather than simply aligned with each other” (Peretz et al, 2004, p. 375). They proved their propositions with the case of people suffering from aphasia without amusia as well as amusia without aphasia. Aphasia without amusia basically refers to the condition where “brain lesion has impaired the processing of language without interfering with the processing of music”, whereas patients of amusia without aphasia are fully functional in language, but cannot recognize music any more (Peretz et al, 2004. p. 4). That is to say the aphasic patients can still sing familiar tunes or learn new ones, but they are not capable of producing intelligible lyrics or utterances in singing or speaking, which further implies that speaking and singing are both controlled by the same language output system (Rebuschat et al, 2011).
Another perspective to explain the close relationship between singing and speaking is speech therapy. Therapies using music such as Melodic Intonation Therapy have been extensively used to cure patients with aphasia. The Melodic Intonation Therapy, originally invented based on the premise that singing can stimulate the right hemisphere of the brain, consists of such exercises as repeating, left hand tapping, intoning and syllable lengthening, and the latter two is comparable to methods used in singing training (Norton et al, 2009). Research has indicated that people suffering from aphasia showed an improved ability to reproduce speech spontaneously and also structural changes in their vocal system after receiving speech therapy, and such changes can be compared to those in vocalists (Schlaug et al, 2009; Halwani et al, 2011, cited in Christiner and Reiterer, 2016). However, it still remains unclear why some people recovered but some did not, and which component of this therapy contribute most in speech recovery. Some researchers (Stahl et al. 2011) proposed that it is rhythm, reflected in left hand tapping, that helps improve speech recovery, while some others (Schlaug et al, 2009) were in favor of the combined effects of all exercises (Christiner and Reiterer, 2016).
2.4 Studies Confirming Musical Aptitude/Singing-Foreign Language Pronunciation Relationship
conducted a research on 118 American college students whom all except two had French in college. The participants of this study took two subtests of the Seashore Measures of Musical Talents: pitch discrimination and tonal memory, while their pronunciation accuracy and accent in foreign language were rated by their teachers. The intelligence and personality of the participants were also included in the test. In pitch discrimination test, the participants had to identify, after hearing two successive tones, whether the pitch of the second went up or down compared to the first one. In the test of tonal memory in which two musical phrases with same rhythm were played, the participants had to determine whether they had the same tonality and point out the pitch of which tone in the second phrase was changed (Jiang, 2005). The results indicated that there was a strong correlation between pitch discrimination and ratings of French accent and intelligence and ratings of French accent, but no relationship between tonal memory and French accent.
Later Leutenegger, Mueller and Wershow (1965), with the use of the subtests of Seashore Measures of Musical Talents, explored the effects of musical aptitude on language learning as well as the effects of language learning on musical aptitude. Two groups of students: one studying French and the other Spanish participated in this study. In terms of musical aptitude, the subjects’ ability to detect difference between each pair of pitch, tone, rhythm and note was tested by the Seashore Measures of Musical Talent, whereas their language learning ability was evaluated by the scores on five tests comprising the Orientation Test Battery of the University of Florida obtained from the Board of University Examiners. These tests include: Committee on Diagnostic Reading Survey Test, ETS English Expression Test and School and College Ability Test. This study, though still only focused on the perception of music and foreign language learning, also explored the differences between genders. No significant correlation between the Seashores subtests and achievement in foreign language learning was found, but the tonal memory scores of the female group predicated the scores in French learning significantly. In all, there was a significant difference between males and females, with females outperforming their counterparts in all aspects, especially in the test of reading.
hearing. The pronunciation of the subjects was tested in the speech production test, and in this test they needed to finish a vowel production task and a random conversation for participants with advanced French levels. The accuracy of pronunciation of the participants were secured by firstly one native French speaker who was invited to write down what vowels they heard from the participants’ recording, and two other native French speakers simply rating on the accentedness of the conversation. In addition to this, the effect of music training on second language learning was also investigated by questionnaire survey. This study ended up with a significant correlation between music perception (rhythm) and speech production (r = .15), music perception (rhythm) and accent production (r = .34) as well as music production and accent production (r = .34). It thus implies that accurate perception is a prerequisite for an accurate pronunciation. However, no significant relationship between music training and second language learning was found.
Slevc and Miyake (2006) examined not only how musical ability was correlated with proficiency of L2, but more variables relevant to the proficiency of L2 such as age of L2 immersion, use of L2, phonological short-term memory and some other factors on 50 Japanese immigrants. These immigrants arrived in America after the age of 11 and spoke Japanese as their native language. They took four tests for their language proficiency, which are receptive phonology, productive phonology, syntax and lexical knowledge. In the productive phonology task, the participants not only had to read 26 minimal-pair words and 26 minimal-pair sentences, but had to read a short English passages. How strong their accents were was judged by two native English speakers. In terms of musical perception, three subtests of Wing’s battery (Wing, 1968) were used, and the participants were asked to accurately repeat three to seven-tune notes immediately after hearing in the musical production task. Further, the participants were also asked to self-rate their own musical ability and singing ability. The phonological short-term memory of the participants was tested by digit span and nonword repetition in which they needed to recall the list of numbers and repeated 40 nonwords of 1 to 4 syllables. The results revealed that musical ability was the strongest predictor of a non-accent speech (r²=0.38), followed by phonological short-term memory (r²=0.30) and language use and exposure (r²=0.29). The age of L2 immersion and self-rated musical ability showed a weaker correlation with the productive phonology. This suggests that musical ability and phonological short-term memory play a very important part in L2 acquisition, while the age of L2 immersion does not affect the acquisition of L2 significantly for adult learners.
analysis and rhythmic memory, while the language aptitude test consisted of sound discrimination (16 test questions), English Grammar test (20 questions), and a language pitch discrimination test in Chinese. The productive musical and language aptitude were examined in Phase II, in which 41 participants imitated 16 Spanish sentences and 14 Korean words and 16 short rhythm patterns from Irish folksongs to test their abilities to imitate rhythm in language and music respectively. Not surprisingly, the results revealed a significant relationship between receptive musical aptitude and language aptitude，r = 0.37 and productive skills in music and language,
r = 0.55, both independent of general intelligence. A significant difference between
males and females was found in receptive language test, with females scoring significantly higher, but differences observed in productive language and music test were not significant.
Nardo and Reiterer (2009) conducted a research on 66 native German speakers to investigate whether musicality as a whole and detailed musicality measures correlate with linguistics abilities, especially L2 pronunciation. In addition to Gordon’s Advanced Measures of Music Audiation consisting of rhythm discrimination and pitch discrimination test, an introspective questionnaire of the subjects’ self-reported abilities relevant to music, including singing capacity, liking for singing, and instrument playing, was also used to secure the evaluation of musical aptitude. The linguistic abilities were tested by means of L2 pronunciation performance test, L2 pronunciation talent test, subtests of Modern Language Aptitude Test and subtest of TOEFL for English grammar. The strongest correlation was found between the score of pronunciation talent and both the liking for singing and self-reported singing ability,
r =.4, followed by the score of rhythm discrimination, r =.24. A significant gender
difference was only discovered in the liking for dancing, which is beyond the scope of music. The conclusion Nardo and Reiterer (2009) drawn from this is that rhythm perception, pitch perception and liking for singing “are the best ingredients” for achieving excellence and expertise in the pronunciation of foreign languages (p. 238).
rhythm and tonality in the receptive musical aptitude test and repeat the tonal and rhythmic patterns in the productive musical aptitude test. In receptive EFL phonological skills test, the subjects were required to match phonemes in groups of words, identify word stress, discriminate rhythm patters in sentences and judge intonation contour and nuclear tones in pairs of words, while what they did in the productive EFL phonological skills test was to read sentences and words specifically chosen based on phoneme, stress, rhythm and intonation. The findings of the study revealed that musical aptitude, as the strongest predictor, aptitude for oral mimicry and musical interest could predict the EFL phonological skills. The age of onset of English learning showed a negative correlation with the EFL phonological skills, while daily use of spoken language did not appear to correlate significantly with EFL phonological production. Moreover, receptive musical aptitude could influence skills of English perception, and productive musical aptitude also had a direct effect on phonological production of English.
Finally, different from all the above-mentioned studies in which musical aptitude or musicality is tested based on its components, namely rhythm, pitch and tonal memory, Christiner (2013) asked the participants not only to complete Gordon’s Advanced Measures of Music Audiation, but repeat melodies in her research on the relationship between singing performance and second language pronunciation. The role working memory played in second language acquisition, especially in pronunciation, was also studied. 41 German-speaking singers were invited to participate into this study. Like other research on the relationship between musical aptitude and language pronunciation, the musical talent of the participants was also measured by Gordon’s Advanced Measures of Music Audiation, and they were also required to sing the well-known song: ‘Happy Birthday’ and repeat a part of an unfamiliar song so as to find out whether singing performance had an impact on the pronunciation skills. Besides reading the standard text used in phonetic research “The North Wind and the Sun”, the participants also imitated sentences in English and Hindi. Further, a digit span task modified based on Wechsler Digit Span (Wechsler, 1939, cited in Christiner, 2013) in which the participants repeat strings of numbers either backward or forwards, and a “monosyllabic non-words with a German-like phonetic quality” task were applied to test the working memory of the participants (Rota and Reiterer, 2009, p. 83, cited in Christiner, 2013). The results of this study is in accordance with those of previous research that the ability to singing was positively correlated with pronunciation of second languages, r =.581 and working memory correlated strongly with performance of singing an unfamiliar song, r =.453 and imitation in Hindi, r =.730. The author then concluded that both singing talent and working memory are predictors of an accurate pronunciation in second languages.
Despite the substantial empirical studies proving the correlation between music and foreign language pronunciation, the following studies said something different.
Thompson (1991) examined the factors which could possibly predict the accuracy of foreign language pronunciation with 10 Russian-born immigrants arriving in the United States at different ages. The accuracy of English pronunciation was evaluated by reading both sentences and passage, but the musical ability was rated by the participants themselves. Eventually, the author did not recognize musical ability as a predictor of foreign language pronunciation accuracy, but ability to mimic was proven to be important in achieving a native-like pronunciation of foreign languages, besides age of arrival and gender.
Similarly, in the study of Fledge, Yeni-Komshian and Liu (1999) on the influence of age on second language acquisition with 240 native speakers of Korean who arrived in the United States at different ages, the subjects self-rated their own musical ability, ability to imitate foreign accent and dialects and ability to remember how English words were pronounced. The three abilities were classified into one category as sound processing ability. Their pronunciation of English was measured by repetition of twenty English sentences, which were later rated by native speakers. However, the results of this study did not reveal a significant correlation between sound processing ability and second language acquisition.
Knickerbocker (2007) conducted a research on 22 people with different educational backgrounds and L1 backgrounds who all studied English as a second language. The tonality subtest of Gordon’s Intermediate Measures of Music Audiation (Gordon, 1982) was used to measure the participants’ musical aptitude. In testing the ability to produce discourse intonation in English, the participants were required to repeat 9 phrases covering 37 English phonemes with the software program Tell Me More. However, it failed to prove the correlation between measures of musical aptitude and discourse intonation.
2.6 Studies on Factors Affecting Foreign Language Pronunciation: Gender, Age of Acquisition and Frequency of Use
Numerous studies have been carried out to explore factors which affect the accuracy of foreign language pronunciation, including age of learning, length of residence in an L2-speaking country, gender, formal instruction, attitude towards language acquisition and motivation, language learning aptitude, pattern of language use and so on (Piske, MacKay and Flege, 2001). Since they are not the major variables investigated in this study, this section will provide a brief review of studies on the role of gender, age of acquisition and frequency of use in foreign language pronunciation.
Piske, MacKay and Flege (2001) conducted a research on 90 people including 18 native English speakers as control group and 72 native Italian speakers who moved to Canada at different ages to explore which factors affected the accent of their English pronunciation, covering gender, age of learning, length of residence, L1 use and L1 ability. All 90 people were required to repeat three English sentences and their performance was rated by nine native English speakers. The results indicated a significant correlation between all factors except gender and accent of English pronunciation, with the strongest correlation found between age of learning and accent of English pronunciation, r= -.83*. However, the authors claimed that age of learning and L1 use could not independently predict the degree of L2 accent due to the possible influence of other intervening factors like motivation, and they also attributed the insignificant difference between genders to a “lack of statistical power” (Piske, MacKay and Flege, p. 209).
Nonetheless, the study of Diaz-Campos (2004) on 46 Spanish learners, which took into consideration the reported use of Spanish both inside and outside the classroom, found a significant difference between males and females in their performance of reading texts with 60 targets words, with females scoring higher in their native-likeness weighting 0.619 in logistic regression analysis than males with a weight of 0.264. The author linked such a finding to the sociolinguistic finding of females’ speech pattern that “female students are more careful in producing the target segments, which can be interpreted as a conservative tendency” (Diaz-Campos, 2004, p. 268). Further, both the reported use of Spanish inside and outside the classroom showed a great influence on the pronunciation, with those using Spanish more sounding more nativelike than those using less, but no further explanation of how the amount of use affects pronunciation was provided.
2.6 Summary of Studies
foreign language pronunciation were reviewed in this section, among which eight acknowledged the existence of such an association and three denied it due to the validity of test instruments or subject sample. Four out of the eight studies confirming the link between musical aptitude/singing and foreign language pronunciation (Morgan, 2003; Slevc & Miyake, 2006; Gilleece, 2006; Pei, 2010) investigated musical production by simply repeating musical notes, intervals and rhythm patterns and only one (Christiner, 2013) integrated singing a relatively long musical piece with lyrics into their experiment. Three of them (Leutenegger, Mueller & Wershow, 1965; Gilleece, 2006; Nardo and Reiterer, 2009) compared the difference between males and females, but only the former two found a significant difference in language production and speech perception respectively and neither of them explained why there would be a difference. Two studies (Sleve & Miyake, 2006 and Christiner, 2013) proved the positive effect of short-term memory and working memory on foreign language pronunciation, and one study (Pei, 2010) confirmed the role of age of English acquisition in EFL phonological proficiency, but denied the relationship between daily use of spoken language and EFL phonological proficiency. The remaining three studies discussed factors that could affect the pronunciation of foreign languages, including working memory (Kormos and Sáfá, 2008), gender (Diaz-Campos, 2004; Piske, MacKay and Flege, 2001), age of acquisition (Piske, MacKay and Flege, 2001) and frequency of use (Diaz-Campos, 2004).
Despite the great attention attached to the link between the music and foreign language acquisition and few existing studies on singing ability and foreign language pronunciation, no study has made a distinction between the innate and acquired singing ability and investigated how singing talent, the innate singing ability, interacts with the pronunciation of foreign languages. Therefore, this study aims to uncover what role singing talent plays in foreign language pronunciation, which probably explains partly why individual differences exist in the attainment of spoke languages
2.7 The Present Study
illustrations (Creswell, 2009). Moreover, the result of a quantitative study can be used to generalize a larger population across the research field. The following research questions are addressed in this study:
1. Is there a correlation between singing talent and foreign language pronunciation?
2. Is there a relationship between working memory and foreign language pronunciation?
3. Is there a difference between males and females in the pronunciation of foreign languages?
Based on the findings of previous research concerning the relationship between singing and second language pronunciation by Christiner (2013), Kormos and Sáfá (2008) and Diaz-Campos (2004), the following hypotheses are made:
1. There is a positive relationship between singing talent and foreign language pronunciation. It means that people who have a better singing performance also do better in the pronunciation of foreign languages.
2. There is a positive correlation between working memory and pronunciation of foreign languages. The longer the working memory span is, the better the pronunciation of foreign languages is.
3. Females and males differ from each other in the reproduction of a foreign language. The expectation is that females are likely to outperform males in spontaneous reproduction of a foreign language utterance.
Chapter 3. Research Methodology
This chapter presents the participants, instruments, procedures and data preparation and processing.
The participants of this study were 41 randomly chosen Chinese people who all speak Mandarin or a Chinese dialect as their native language, as a sample of at least 30 participants can achieve a normal distribution (Field, 2009, cited in Christiner, 2013), but 10 of them were excluded for the final analysis because they either failed to follow the instructions or did not finish all the tests. The rest of the participants consists of 14 males and 16 females, with a mean age of 23.93 (range 18-33). Some of the participants (9) are pursuing or have obtained a master degree in China or in the Netherlands, 9 of them are freshman in a normal university in China and the rest (12) graduated with a bachelor degree. All the participants speak English as their second language and English is also the foreign language that all participants are proficient in, except one female speaking Korean as her second language and one male capable of speaking French and Japanese. None of the participants have received training in music or singing for more than half a year, and two of them are able to play an instrument, but they only practiced for just a few months in their earlier years. Such a short period of musical training is generally considered as no musical training in the studies reviewed in the previous chapter.
3.2 Instruments
Five instruments were applied in this study, including a self-designed questionnaire, an English reading task, an English imitation task, a Catalan imitation task, an English song singing task and an auditory digit span test provided by millisecond.
3.2.1 Questionnaire
The questionnaire was divided into two sections, the first of which covers the general information of the participants including their initials, age, gender, educational background, mother tongue, language they know, age of English acquisition, musical experience, handedness and hearing. The second section consists of the participants’ self-rating of their singing ability, English pronunciation accuracy and accent imitation ability, their frequency of English usage in daily life and whether they like music and singing. The self-rating is given on a scale of 1 to 5, ranging from ‘very bad ( recorded as 1)’, ‘bad (2)’, ‘fair (3)’, ‘good (4)’ to ‘excellent (5)’. The complete copy of this questionnaire together with the instruction of other tasks can be found in Appendix 1.
3.2.2 English Reading Task and Sentence Imitation Tasks
participants’ pronunciation skills.
Firstly, the Aesop’s Fable “The North Wind and the Sun”, which is the standard text used in phonetic studies, was used in the English reading task to test the participants’ proficiency in spoken English. The following paragraph is used in this study.
The North Wind and the Sun were disputing which was the stronger, when a traveler came along wrapped in a warm cloak. They agreed that the one who first succeeded in making the traveler take his cloak off should be considered stronger than the other. Then the North Wind blew as hard as he could, but the more he blew the more closely did the traveler fold his cloak around him; and at last the North Wind gave up the attempt. Then the Sun shined out warmly, and immediately the traveler took off his cloak. And so the North Wind was obliged to confess that the Sun was the stronger of the two (International Phonetic Association, 1999, p. 44).
Next, the participants’ foreign language pronunciation skills were furthered measured by two spontaneous sentence imitation tasks, one in English with uncommon words and the other in Catalan, a language none of the participants were familiar with. Since Catalan was a language that all the participants have never heard or been exposed to before, eight short and simple Catalan sentences, including six narrative sentences and two interrogative sentences were chosen for this task to test the participants’ ability to repeat an unfamiliar language under the suggestion of a native Catalan speaker. The original Catalan sentences, serving as the example for the participants to imitate, were recorded by a native Catalan speaker with cellphone. The sentences used in this task are as follows:
1. Estem buscant feina. 2. Teniu gana ara.
3. Són les vuit del vespre. 4. Casa seva és molt gran. 5. Què li passa a aquest paio? 6. Volia anar a Espanya. 7. La meva tieta és mestra. 8. Qui sou tots vosaltres?
that all the participants are most proficient in, except for one whose second foreign language is Korean, the participants’ ability to imitate and pronunciation skills will not be reflected in repeating sentences with simple words. Therefore, each of the English sentence contains one to two words that are considered as the most difficult English words to pronounce for non-native speakers from Netizens on Reddit, which is a website of social news, web content rating and discussion with monthly visitors of 542 million (Wikipedia, 2017). Similar to the Catalan sentences, the original English sentences, recorded by a native British English speaker with cellphone, were played to the participants as examples. The following sentences were used in this task.
1. Do you like Worcestershire sauce? 2.There is miscellaneous information. 3. She is a rural juror.
4.Here is the sixth sheep.
5. Look at that mischievous squirrel. 6. I am suffering from mild pleurodynia. 7. Never mind the triskaidekaphobia. 8. It takes up a brobdingnagian proportion.
3.2.3 Singing Talent Test
As the aim of this study is to explore the relationship between singing talent and foreign language pronunciation, the test of singing talent is of crucial importance. On the one hand, the repetition of a short simple melody or a basic tune is manageable for almost everyone, as people have the basic abilities to repeat familiar or simple songs. On the other hand, learning and then singing a new song including melody and lyrics allows for the evaluation of more sub-abilities like working memory (Larrouy-Maestri et al, 2013; Christiner and Reiterer, 2013). After several pilot tests, the song “You Belong to Me” sung by Carla Bruni, which has fairly simple rhythm and melody and none of the participants had heard of it before, was adopted to measure the singing talent of the participants. This song lasts for three minutes and one second in all, but the participants only needed to repeat the first 52 seconds, including the following lyrics:
Just remember darling all the while You belong to me
See the marketplace in old Algiers Send me photographs and souvenirs Just remember till your dream appears You belong to me.
The participants’ singing performance was recorded by a Chinese singing APP named 全民 K 歌 (WeSing) and automatically rated by this APP based on melody, rhythm, and intonation. As the validity of the rating system of this APP has not been academically proved, the singing performance was also graded according to melody, rhythm and intonation as well by five graduates from Beijing Contemporary Music Academy who have been trained in singing performance for at least 10 years.
3.2.4 Working Memory Test
According to Rota and Reiterer (2009), phonological working memory can predict the pronunciation skills in foreign languages. Thus an online auditory digit span test
provided by millisecond, obtained from
http://www.millisecond.com/download/library/digitspan/, was used to test the working memory of the participants. The test is divided into forward task and backward task, but only the forward one was used in this study due to the limit of time. In the test, the participants heard a sequence of numbers from 1 to 9, then chose the numbers they heard in the same order as given. The span of numbers presented could range from 3 to 15. The working memory span of the participants is the maximum string they could repeat correctly. The final results consist of maximum number of digits recalled correctly and maximum number of digits recalled correctly before making two consecutive mistakes, but only the former one was used for later analysis.
3.3 Procedures
This study was carried out in March, 2017, but not all the participants were met face-to-face due to different residential regions. Nine of them were tested by the researcher in person, twenty of them were tested by a friend of the researcher face-to-face,and the rest (11) did the test by themselves under the guidance of the
informed of the objective, testing procedures and estimated time needed for the study and ensured of the confidentiality and anonymousness of their answers in the questionnaire survey and performance in other tasks so that they could be more cooperative and supportive. The entire process of the experiment is shown in Figure 1.
Figure 1. Flowchart summarizing the entire experiment process.
3.3.1 Questionnaire Survey
The questionnaire survey was carried out with Qualtrics. Before filling out the questionnaire, the participants were told how to operate with their computer. While answering the questionnaire, the participants were allowed to ask questions and explanations for items they were unclear of. After finishing the questionnaire, the participants then submitted their answers.
3.3.2 English Reading and Sentence Imitation Tasks
The second task in this study was reading “The North Wind and the Sun”. Before being recorded, the participants were given some time to get familiar with the text and practice for at most five minutes. There was no restriction on which accent they should sound like. In the two sentence repeating tasks, each of the sentence recordings was played three times to the participants, who started repeating immediately after the third time, and then proceeded to the next sentence. The recording was done with “Voice Memo” function of an iPhone6 and Huawei P9 and sent to the researcher and raters later.
Five native English speakers with the qualification of teaching English to speakers of other languages including two from Australia, two from the United States and one from UK were involved in rating English imitation. They were told that they need not focus on a specific accent (see Appendix 2 and 3 for details). However, as it is
Questionnaire English
reading
difficult to find Catalan-speaking specialists to get engaged in rating the Catalan sentences, five ordinary native speakers with an educational background of bachelor and above were invited based on the finding that there is a significant agreement on ratings of foreign languages based on intelligibility, nativeness of pronunciation and so on between both linguistically experienced and inexperienced raters (Calloway, 1980), whereas Cartier (1975) speculated that those who are skilled in understanding speech with different accents such as language instructors could be more tolerant than ordinary people in accepting accent. The raters rated the recordings on a scale from 1 (little resemblance to a native speaker) to 10 (perfect native-like) based on the intonation, word stress, rhythm, intelligibility and pronunciation accuracy, and half score can also be used (full instruction can be found in Appendix 4). However, due to personal reason, one Catalan rater did not manage to finish rating the recordings of all the participants, and only the scores from four raters were used in later analysis. In order to maintain an equivalence between the results of English and Catalan task, the result of English task, which is least correlated with the others was excluded.
3.3.3 Singing Talent Test
The participants first listened to the song “You Belong to Me” sung by Carla Bruni three times and started to sing immediately after the third time with the built-in microphone of earphones of iPhone 6 and Huawei P9, as not wearing the earphones could result in the capture of sound from accompaniment, affecting the ratings. The score of the participants’ singing performance was presented right after they clicked on “Stop recording” and they made a screenshot of the scores, uploaded the recording and privately shared the link of the recording with the researcher so that the latter could download it for further rating. The evaluation on singing performance was conducted based on intonation, rhythm, quality of voice and melody ranging from 1 to 10, where 1 equals little accuracy compared to the original performance and 10 means a perfect and accurate representation of the original piece (full instructions can be found in Appendix 5).
3.3.4 Working Memory Test
3.4 Data Preparation and Analysis
The information collected in the questionnaire was transformed into nominal variable (gender, educational background), ordinal variable (frequency of English use, musical interest and fondness of singing) and interval (age, age of English acquisition, self-rated English pronunciation, self-rating singing ability, self-rated ability to imitate a strange accent) and input into an excel file.
Since there were 30 English reading files and 240 files (8 sentences and 30 participants) of English sentence imitation for English raters and 240 files of Catalan sentence imitation for Catalan raters, it would be too tiring for raters to finish rating with online instruments such as SurveyMonkey without a break, and they would become impatient or be tolerant in their ratings later after hearing the same sentences again and again. As a result, all the audio files and a word file including rating criterion (see Appendix 2) were sent to the raters via email, so were the recordings of singing task.
The total score the participant could get was obtained by taking the mean value of the sum from all raters. The maximum score for English reading task is 50, 400 in English sentence imitation, 400 in Catalan sentence imitation and 40 in singing. An outlier was found in the final score of English reading task, but it was not excluded as it does not affect the result. After calculation, all scores were merged into one excel file together with digit span of working memory test and variables related to the research questions of the study.
SPSS Statistics Version 19.
Before examining the correlation between singing and foreign language pronunciation, a Pearson r correlation test was first run to check the validity of the rating system of “WeSing” App, which had never been academically verified. The result showed that the scores given by the App were strongly positively correlated with those by raters,
r= .886, p<.001 (two-tailed). Clearly, the rating system of the App is valid, but the
singing scores given by raters would still be used in later calculation and analysis, as the performance of each language imitation task was also graded by human. Detailed correlation data between scores of singing given by the App and foreign language pronunciation can be found in the Appendix 8.
Chapter 4. Results
This chapter presents the results of the tests in response to the research questions of this study. Some extra-linguistic variables included in the questionnaire were also examined. The Shapiro-Wilk test, which is used for normality test when the sample size is smaller than 50, showed that scores of English reading and digit spans were not normally distributed, and therefore, both parametric and non-parametric tests were used depending on the dataset.
4.1 Correlation between Singing and Foreign language Pronunciation
The first research question was: Is there a correlation between singing talent and foreign language pronunciation? The expectation was that there would be a significant positive correlation between them.
Pearson r correlation tests were conducted respectively between singing performance and English imitation task, singing performance and Catalan imitation task and Spearman rho test for singing performance and English reading task. The results of the correlation between singing and language pronunciation can be found in Table 1.
Table 1. Correlation between singing and pronunciation
English imitation Catalan imitation English reading
p < .001 p < .001 p < .01
**Correlation is significant at the 0.01 level (2-tailed).
The results indicated a strong positive correlation (when .40 < r < .69) between not only singing and spontaneous language production tasks: singing and English imitation, r = .654**, singing and Catalan imitation, r =.540**, but singing and practiced English reading r =.483**, though the correlation was the weakest among all three language pronunciation tasks. Since the p values were all smaller than the α value of 0.05, the hypothesis could not be rejected. It then provides evidence to the hypothesis of the strong positive correlation between singing talent and foreign language pronunciation. The better one did in singing, the better he/she was in achieving an accurate and native-like pronunciation. The correlation between singing task and English reading task, English sentence imitation task and Catalan sentence imitation task were shown respectively in Figure 2, 3 and 4.
Figure 3. Correlation between English imitation and singing
Figure 4. Correlation between Catalan imitation and singing
4.2 Correlation between Language Pronunciation and Working Memory
was that there would also be a positive relationship between working memory and foreign language pronunciation.
Spearman rho tests were run, for not all datasets follow a normal distribution. The results in Table 2 showed that working memory was strongly positively correlated with not only all language production tasks but singing with the strongest correlation,
r = .919. This suggests that the longer the digit span, the better the spontaneous
reproduction of English and Catalan and singing. The correlations between working memory and language pronunciation are respectively presented in Figure 5, 6 and 7.
Table 2. Correlation between working memory and language pronunciation
English
imitation Catalan imitation English reading Singing WM r = .648** p < .001 r = .587** p < .005 r = .482* p < .01 r = .919** p < .001
** Correlation is significant at the 0.01 level (2-tailed). * Correlation is significant at the 0.05 level (2-tailed).
Figure 6. Correlation between English imitation and working memory
Figure 7. Correlation between Catalan imitation and working memory
4.4 Subjects Divided by Gender
This last research question is whether there is a difference in foreign language pronunciation between males and females, with our expectation affirming the existence of difference.
found between two groups in a Mann-Whitney Test, as the scores for females did not follow a normal distribution. In English sentence imitation task, a significant difference was found between two groups, t(28) = 2.51, p < 0.05. Females had a mean score of 281.05, SD =31.90, while males had a mean score of 253.77, SD = 28.14. In Catalan sentence imitation task, females (Mean =189.36, SD =45.20) also outperformed their counterparts (Mean =152.93, SD =30.07). There was a significant difference between two gender groups, t(28) =2.56, p < 0.05. In terms of singing, the mean score of females (n =16) was 23.30, SD =4.85, and that of males (n =14) was 15.70, SD =5.96. An independent-samples t-test showed that there was a significant difference between the two gender groups, t(28) =3.84, p < .001 (two-tailed). These results indicated that females generally could do better in singing and foreign language learning. The means of both gender groups in the four tasks are presented in Figure 8.
Figure 8. Means of females and males in four language production tasks
4.5 Other Variables in the Questionnaire
accent and self-assessment of singing than males. Other than that, females were generally more interested in music and fondness of singing. However, a significant difference between males and females was only found in the age of acquisition, t(28)= -2.426, p < 0.05 and fondness of singing, t(28) =2.385, p < 0.05 according to the independent-samples t-test.
(Notes: SAE=self-assessment of English pronunciation, SAI=self-assessment of imitation ability, SAS=self-assessment of singing, IM=interest in music, FEU=frequency of English use, AoA=age of English acquisition, FoS=fondness of singing)
Figure 9. Means of females and males in the seven variables in the questionnaire
Table 3. Correlation between each variable and language pronunciation
Independent variables English imitation Catalan imitation English reading
Age of English acquisition r = -.188 p > .05 r = -.153 p > .05 r = -.092 p > .05 Self-rated English pronunciation r = .451** p < .05 r = .485** p < .01 r = .227 p >.05 Self-rated imitation ability r = .375** p < .05 r = .420** p < .05 r = .226 p > .05
Self-rated singing ability r = .452** p < .05
r = .457** p < .05
r = .594** p < .05
Frequency of English use r = .455**
p < .05 r =.219 p > .05 r = .422** p < .05 Fondness of music r = .189 p > .05 r = .146 p > .05 r = .242 p > .05 Fondness of singing r = .413** p < .05 r = .360 p > .05 r = .408 p < .05 **Correlation is significant at the 0.01 level (2-tailed).
Finally, two multiple regression tests were run to examine the variables that contributed most to the variance in language pronunciation in two spontaneous language production tests. All the variables investigated in this study were entered for multiple regression analysis, but only singing talent measured by singing performance, self-rated English pronunciation and frequency of use were left in the test for English imitation task and working memory and self-rated imitation ability for Catalan imitation task, and all the other variables were ruled out.
Catalan pronunciation.
Table 4. Multiple regression indicating the strongest indicators for pronunciation in English imitation
Step Variable R² Adjusted R² F value p value
1 Singing performance 0.427 0.407 20.885 .000
2 Self-rated English pronunciation 0.579 0.547 18.529 .000
3 Frequency of English use 0.655 0.616 16.482 .000
Table 5. Multiple regression indicating the strongest indicators for pronunciation in Catalan imitation
Step Variable R² Adjusted R² F value p value
1 Working memory 0.385 0.335 15.613 .000
2 Self-rated imitation ability 0.488 0.450 12.873 .000
Chapter 5. Discussion
In this chapter, the results presented in the previous chapter are discussed. This study mainly explored the relationship between singing talent and foreign language pronunciation. Other possible factors that are likely to result in a difference in achievement in foreign language pronunciation including working memory, gender, age of English acquisition and frequency of English use pertinent to English pronunciation are also taken into consideration. The three research questions regarding the relationship between the pronunciation of unfamiliar words in English and a completely unknown language (Catalan), the role of working memory in foreign language pronunciation, gender difference in the performance of foreign language pronunciation are addressed and discussed next. The age of English acquisition and frequency of English use related to English pronunciation is also briefly discussed.
5.1 Singing and Foreign Language Pronunciation
expected that there would be a positive relationship between singing talent and foreign language pronunciation, which is also proved by the results as indicated in the previous chapter. The first hypothesis that there is a positive correlation between singing talent and foreign language pronunciation accuracy was, therefore, attested. This finding was consistent with that in the studies of Morgan (2003), Slevc & Miyake (2006), Gilleece (2006) and Pei (2010) whose focus was on musical aptitude measured by reproduction of short music patterns and foreign language pronunciation, as well as Christiner (2013) who tested the singing ability on professional vocalists. The strong correlation between singing performance and pronunciation of foreign languages suggested that singing talent of people receiving no training could be a good predictor for the pronunciation of foreign languages, and good singers have advantages over poor singers in pronouncing foreign languages, no matter with which they are familiar or not. Also, the strong positive correlation between the self-reported fondness of singing and pronunciation found in this study was consistent with that in Nardo and Reiterer (2009), who concluded the liking for singing as one of the three most influential factors for successful foreign language pronunciation.
Such an association between singing and foreign language pronunciation can be firstly accounted by their nature that they share the same vocal system and part of neural mechanisms and both require certain motor skills (Christiner, 2013). Human has two pathways for vocal learning like songbirds, one of which is the “striatal vocal learning pathway”, responsible for learning speech, and the other is the motor pathway, responsible for the reproduction of speech learned previously (Jarvis, 2004, as cited in Simmonds, 2015, p.1). Simmonds (2015) hypothesized that the motor pathway of late foreign language learners became inactive, resulting in a non-native accent, but it can become active again by practicing. On the other hand, singers or people talented in singing generally have a more flexible and active vocal motor ability, involving the proper use of vocal organs, such as lips, tongue, larynx and jaw. This ability can be compared to the flexibility of infants that allows them to acquire not only familiar languages and phonemes but those they are never exposed to within in a certain period (Murphey, 1990, cited Christiner and Reiterer, 2013). Since foreign language learners need to repeat again and again before they can finally coordinate how their vocal organs work to make a correct and native sound, people good at singing, knowing how to articulate flexibly with a better manipulation of their vocal organs, certainly can adapt to the pronunciation of foreign languages better and faster.
others believe that a poor imitative ability can also be responsible for an inaccurate pronunciation (Purcell and Sutter, 1980). According to Graham (1969, cited in Morgan, 2003, p. 33), “the transference of varying degree” of the music of mother language to the foreign language can be a cause of accent in the foreign language, resulting from a lack of abilities to imitate. Several studies such as Purcell and Sutter (1980), Thompson (1991), Morgan (2003) and Christiner and Reiterer (2013) have also recognized the importance of mimicry ability for the accuracy of foreign language pronunciation.
More evidence of the close relationship between singing and the process of language acquisition can be found in how native language is acquired. When mothers or caretakers speak to infants, their utterances are more song-like with more pauses, pitch variations, slower and simpler so that infants can easily recognize and imitate them (McMullen and Saffran, 2004; Kuhl et al, 1997). In earlier months, infants are capable of learning all languages with their openness to all phonemes (Kuhl, 2004), but such plasticity disappears as they have more control over the vocal motor system to produce language features of their mother tongue (Iverson, 2010, cited in Christiner and Reiterer, 2016). Furthermore, infants generally are more sensitive to the intonation, tonal and rhythmic properties of their mother tongue, and acquire the language by repeating prosodic components, to be specific, the intonation of discourse at first and develop the communication skills gradually through interaction (Mora, 2000).
Another aspect to interpret the relationship between singing and foreign language acquisition is the use of singing as a teaching instrument in foreign language classes. This, however, has little to do with singing talent, and is not specific to phonetic acquisition only. Singing, on the one hand, can make it easier to memorize not only pronunciation but words and expressions, and reduce the fear of learning and speaking an unfamiliar language, probably because it is affective and less energy-demanding (Mora, 2000; Christiner, 2013). On the other hand, it can enhance learners’ awareness of rhythm and intonation, two important components in spoken languages (Mora, 2000), and thus they can adapt to unfamiliar rhythm and intonation more easily. Just as what Techmeier (1969) demonstrated in the study on American students who made an improvement in the pronunciation of French after joining the choir, “singing tunes the ear to the delicate shades of sound besides promoting articulation of words and focusing attention on proper accentuation” (p.96, cited in Morgan, 2003).
