Closing Diphthongs among Javanese Learners of English

This paper describes adult acquisition of L2 English diphthongs among Javanese learners of English. The objectives of this study are concerned with whether vowels are more or less diphthongal by measuring the change of first formant (F1) and speech duration to examine the influence of Javanese (L1) phonetic system on the acquisition of English as a second language. The acoustical analysis was conducted by comparing the formant frequencies and duration of L2 with the corresponding diphthongs by native speakers of English (NE).The purposewas to test the hypothesis that difficulty in acquiring an L2 contrastive category is related to the role in the L1 of the phonetic feature upon which the L2 category is based. Although F1 values of L2 differed in the start point of closing diphthongs, there was only one closing diphthong /QŠ/ produced by L2 which was statistically different from L1 based on the F1 changes. Surprisingly, the diphthong /Tj/ showedan increase F1 value instead of decrease which means that the Javanese production was more open. The result in duration showed that there were crucial differences in diphthongal duration produced by Javanese learners of English. Javanese tended to lengthen the diphthongal words.


INTRODUCTION
In English as second language (ESL)teaching, learners of English particularly adult learners, frequently make mistakes even failure in producing native like pronunciation because their L1 highly influences L2 production.Admittedly, as a speaker of a language gets older, mother tongue interference cannot simply be dislodged.Adults tend to stabilize their language learning at a certain stage.Ellis (1994, p.09) stated that learner speech was 'structurally organized' in the sense that it constituted a system in its own right.This process called fossilization; when a sound in L2 consistently replaced with a sound which is phonetically close to L1.Thus, the speech production in L2 is different from the similar speech produced by a native speaker.Selinker (1972) also noted that relatively few adult learners reach native-speaker competence.Contrastive analysis hypothesis (CAH) proposed by Lado (1957, p.2) was also based on the assumption that people who comes into contact with a foreign language will find some features of it quite easy and others extremely difficult.Those elements that are similar to their native language will be simple for him, and those elements that are different will be difficult.
Likewise, Javanese learners of English frequently have difficulties with uncommon English vowel system because there are dissimilarities in articulating vowel sounds of these two sound systems.In L2 learning, the level of difficulty experienced by the learner will be directly related to the degree of linguistic difference between L1 and L2.Difficulties will manifest themselves in errors; the greater the difficulty, the more frequent the errors (Ellis, 1994, p.308).Similarly, the difference between English and Javanese vowel system could be a barrier for Javanese learners of English who want to sound more native-like.
Javanese and English have vast differences in vowel system particularly diphthongs.Javanese phonology includes 8 vowel phonemes: 6 phonemes and 2 allophonic pairs (Uhlenbeck, 1963).An Indonesian linguist, Marsono (1999, pp.55-58), reported 10 vowels (including allophonic variants) and 5 diphthongs in Javanese.English, on the other hand, has a large number of diphthongal sounds which consist of a movement or glide from one vowel to another.One of the most common pronunciation mistakes that result in a learner of English having a "foreign" accent is the production of pure vowels where a diphthong should be pronounced (Roach, 1998, p.18).
Each vowel including diphthong has a unique structure of formant frequency which indicates the quality of the vowel.Formants have long been known as suited parameters for describing vowel production due to its correlation with traditional articulatory transcription of vowel (Hawkins & Midgley, 2005).The relationship between Fl and F2 determines the acoustic quality of the vocoid articulations.The longer the cavity, the lower Fl will be; the shorter it is, the higher Fl will be.This correlates with tongue-height; the pharyngeal tube continues into the mouth and is narrowed in varying degrees by the movement up and down of the tongue.Similarly, F2 correlates with the length of the oral cavity in terms of frontness and backness of the tongue body.The longer the front cavity, the lower F2 will be (= back vowels); the shorter it is, the higher F2 will be (= front vowels) (Lodge, 2009, pp. 190-199).
Thus, formant frequencies can be used to measure the accuracy of vowel production.Nevertheless, in Indonesia, English pronunciation instruction mainly focuses on audio-lingual approach such as minimal pair drills and articulatory descriptions.Correction and analysis of students' production generally depends on the teachers' listening judgment.In fact, one of the accurate ways to discern the differences would be based on acoustic analysis.Therefore, this research attempts to investigate L2 production in English pronunciation especially diphthongs by examining the acoustic difference between L2 diphthongs and native speakers (NE) diphthongs.
This study investigates pronunciation of L2 diphthongs and attempts to answer two questions ( 1) to what extent is the pronunciation of L2 diphthongs different from the corresponding diphthong in NE? (2) is there any difference in speech duration between L2 and NE.

PREVIOUS STUDIES
The previous studies on L2 speech production in Indonesia have scarcely been done and mainly based on auditory judgment and experience of teachers or researchers in teaching practice.Many scholars have described the characteristics of English pronunciation produced by Javanese learners.Nonetheless, studies about Javanese learners of English particularly in acoustical phonetics have rarely been conducted by Javanese or Indonesian linguists.In Indonesia, studies about non-native pronunciation are mainly based on phonological interpretation such as minimal pairs and listening judgment.There are only few researches related to the study.Perwitasari (2015) writes about vowel durations in English as a second language among Javanese Learners.She finds that Javanese Learners of English seemed unaware of long and short duration of English vowels.Another Indonesian researcher, Widagsa (2015) found that Indonesian learners of English are strongly influenced by their mother tongue.Indonesian learners of English make a good impression only when the vowels in English are similar to Indonesian.It is proven by formant frequencies which are in common.
In Singapore, Deterding (2007) conducted a research related to speech production measuring Singaporean English (SE) Diphthongs.He outlined the distinctive characteristics among the Singaporean speakers of English compared with RP.He found that Singaporean speakers are less diphthongal than the corresponding vowels in standard British English.The use of a relatively monophthongal realization of the diphthongs by Singaporeans might be regarded as a distinctive characteristic of the local speech.Different results came from Tsukada (2008) who found that the production of English diphthongs by Thai learners of English were much closer to NE.Further, their production of the diphthongs might have been related to a large number of diphthongs in the Thai vowel inventory, which might have encouraged its speakers to substitute existing L1 categories for the English diphthongs rather than forming authenticnew phonetic categories.
According to the previous explanation, in Indonesia, research related to acoustical phonetics especially in diphthongs of English as a second language have rarely been done.English diphthong of Javanese learners is an interesting topic to study for researchers.Thus, it can be said that this research is relatively new and needs to be conducted.This research will give novel contribution to ELT and other second language learning in Indonesia particularly in local language interference such as Javanese.

ENGLISH CLOSING DIPHTHONGS
Most varieties of English have several diphthongs.The most obvious diphthongs are the vowels of choice, mouth and price in most standard varieties of English.These diphthongs start with open vowels and then rise to close vowels, gener-

JAVANESE DIPHTHONGS
Javanese, one of the most widely spoken languages in Indonesia, is a mother tongue of more than 75 million people (Crystal, 1997;Edi, et al., 2007).It is spoken mainly in Central Java and East Java.Javanese includes 8 vowel phonemes: 6 phonemes and 2 additional allophonic pairs Related to diphthongal vowels, Marsono (1999, pp. 54-58) claimed that Javanese has at least 5 diphthongs, one rising diphthong [ui] as in the words uijo 'extremely green', cuilik 'very small', uireng 'totaly black' and the others are falling diphthongs.

SUBJECTS
This experimental research involves five female subjects who were native speakers of Javanese aged 17-23.They used mainly Javanese in daily communication and lived in Yogyakarta.In addition, they had never been to English speaking countries.The subjects were all students of English Department of Universitas PGRI Yogyakarta (PBI-UPY) and all subjects had been studying English for at least 3 years.To ease the training process, the subjects were informed that the recording of their voice would be used for educational and experimental purposes.The audio data of English native speakers were taken from text-to-speech softwares.

STIMULI
All subjects produced a set of target words.The target words comprised eight diphthongal words such as pay, five, home, now, join, near, hair, pure which were inserted in a carrier sentence"I say (diphthongal words) again".During the recording, subjects repeated the sentence twice.The particular fillers or lexical sets were chosen to distract speakers' attention from the experimental wordsin order to encourage natural pronunciation.

RECORDING PROCEDURE
Before the recording process began, first, each subject completed a consent form, a briefquestionnaire, which provided information about thesubjects' native language and second language background.Second, subjects received a short introduction monologue which contained words simulated for the recording.In order to get accustomed to the target stimuli, they were given time to read and practice.
Subjects were familiarized with the experiment and the procedures of recording.Each subject took a seat in front of a computer screen with active mode recording tools (audio recorders, and microphone).Once the stimuli appeared on the screen, subjects started to produce the sentences.The stimuli are presented in random order.Speech production were recorded in a sound-attenuated room and stored on a computer.Audio recording is treated confidentially and used for acoustic analysis.Last, the recorded speech sounds were annotated and segmented.
The recordings were made with ASUS X200MA notebook and a microphone placed at the distance of about 10 cm from the subjects' mouths.The subjects were recorded one by one in order to ease the analysis.This process took place in the language laboratory of Universitas PGRI Yogyakarta.

ACOUSTIC MEASUREMMENT
Closing diphthongs are sounds which the vowel quality changes from relatively open position to a more closing position during the course of the vowel (Deterding, 2007).As the vowel quality is becoming less open, the F1 at the start point is expected to be different from the end point.The decrease of F1 would be an indication of how diphthongal the sounds are.
The recordings were analyzed using PRAAT 5.3.51.The software allows some features such as tracing the formant frequencies, choosing time point, and drawing waveform and spectrogram display.

STATISTICAL ANALYSIS
To measure a difference between the production of L2 diphthongs and L1 English, this research conducted statistical analysis.An independent t-test for the groups was applied to test whether frequencies and speech duration was significantly different between groups.

FORMANT FREQUENCIES
Diphthongs are the monosyllabic vowels which have two discernibly different points, one at the start and one at the end.Therefore, the soundwave data from the respondents are carefully analysed by defining the two target points.The first target is the starting point (vowel start) and the second is the ending point (vowel end).In measuring diphthongs, the vowel quality of vowel starts and vowel ends must be measured then both vowel qualities will be plotted in a vowel chart in order to figure out the movement of the vowels.
The complete result of the the acoustic measurement of L2 and L1 English closing diphthongs is shown in the Appendix.Table 1 describes the mean of formant frequencies of English closing diphthongs and duration by Javanese Learners and Table 2 shows the mean of formant frequencies and duration of L1 English closing diphthongs Based on the start points, several of the English closing diphthongs produced by Javanese learners of English were nearly identical to NE in regard to vowel height.To ease the comparison, the numerical data are plotted to bar chart.
The following bar chart illustrates the comparison of F1 value of English L2 and NE. Figure 4 demonstrates the start points of F1 while Figure 5 shows the end point fo F1.It can easily be noticed from the previous tables and figures that the most similar result emerged in the pronunciation of /YŠ/(F1 start NE 629,6 Hz and F1 start English L2 598 Hz).On the other hand, the most distinctive diphthong produced by English L2 was /Tj/ (F1 start NE 403 Hz and F1 start English L2 600 Hz).
Unlike the start points, the average F1 value of diphthong /Tj/ was close to the NE production (F1 end English L2 490,6 Hz and F1 end NE 503,2 Hz).Even so, the most disparate F1 end value was OEj with 592,8 Hz for English L2 and 802 Hz for NE.
To observe the decrease and increase of F1 value of English L2, the changes of F1 are presented in the following chart.

STATISTICAL RESULT
Closing diphthongs in English are /ej/,/OEj/,/ YŠ/,/QŠ/,Tj/ and expected to have decrease in F1.In general, the English closing diphthongs produced by English L2 were pronounced near NE in regard to the vowel height.In addition, the statistical analysis also confirms similar result.The following table shows the statistcal analysis employed in this experiment.
Table 2. indicates that there is only one diphthong produced by L2 which is statistically different from L1.The result of the measurement of / QŠ/is higher ([t(8)

VOWEL CHART
In this case, since quadrilateral vowel chart requires the frequency of both formants (F1 and F2), the F2 is used to plot the numerical data Bark Scale.Thus, the numerical data are then plotted in a vowel chart (Bark Scale) to figure out how diphthongal they are.The following figure is the Bark Scale of average formant changes by respondents in producing English closing diphtongs.The greendots indicate the starting points of diphthongs and the yellow square dots represent the ending points of the diphthongs.It can clearly be seen that there are variations among the diphthongal sounds.It can clearly be seen in Figure 5 that the diphthong /Tj/ was produced differently.The arrow points at down left side of the graph.It means that the end point of the diphthong was pronounced as low vowel with half open mouth.

DURATION
In this research, the duration was only measured in the diphthongal words.The statistical analysis of L1 and L2 duration in producing diphthongal words are served in the following table.
Table 2. shows that there are significant difference in L2 duration.The differences are in the / ej/, /QŠ/ and /Tj/.It clearly indicates that L2 has longer duration in pronuncing diphthongal words than L1.The following figure demonstrates the difference.Figure 9 shows that L2 tend to lengthen the duration particularly in /ej/, /QŠ/ and /Tj/.However, there was only one diphthong which was pronounced shorter than native speakers of English; the diphthong /OEj /.

CONCLUSION
Javanese Learners of English do not have references to the set of diphthongs attributed to their first language phonetic features.Javanese are predicted to produce English diphthongs differently compared to native speakers of English.There are several findings which can be drawn based on the results.Admittedly, English closing diphthongs produced by L2 were not statistically different from L1, the different was only on the production of / QŠ/.It should be noted that there is a difference in the change of F1 values of the sound/Tj/.The differences answer the research question that the production of English diphthong by Javanese learners of English was different from the corresponding sounds produced by English native speakers.The other answered question was duration.It shows that there are significant differences in diphthongal duration produced by Javanese learners of English.They tend to lengthen the diphthongal words, three main findings stand out.
To put differently, the results of this measurement particulartly in English closing diphthongs and the duration, support the hypothesis' prediction that the L2 failure in producing L2 vowels is related to the absence of diphthongal sound as a contrastive feature in L1.Javanese does not include diphthongs in its phonological system.It proves that the absence of those features in L2 can be a barrier for learners to study the language.The measurement results highlight a potential learning prob-lem for Javanese learners of English.They are heavily influenced by their mother tongue.
This research is only a preliminary study which still requires further research related to acoustic features of English vowels and consonants produced by non-nativespeakers particularly Indonesian speakers or Indonesian regional language speakers.In this paper, I only report data of male respondents.Age is known to be an important factor of phonetic variation, aged-related would also be interesting topic for future research.Hopefully, this research would be worthwhile for language teaching in Indonesia.
ally in the area of[i]  or[u].These are called closing diphthongs for this reason.Diphthongs are transcribed by the start and end points.For example, the vowel of choice is transcribed in RP as [Ti]: it starts with [T] and ends with[i].Ogden (2009,  pp.70-71)  reports eight diphthongs in British English RP and divides in two main types; RP closing diphthongs and RP centering diphthongs.

FIGURE
FIGURE 1: CLOSING DIPHTHONGS [e] -[å] and [o] -[T].More recent studies confirm that Javanese vowels are grouped into 6 phonemes, including 4 allophonic pairs [i] -[I], [u] -[Š], [e] -[å], and [o] -[T].The allophones of each vowel frequently occur in closed syllables.The standard Javanese of central Java is typically characterized as having six vowel phonemes.In the six vowels system, which appears to characterize the speech of our consultants, [å] and [T] are in complementary distribution with [e] and [o], respectively.

FIGURE 6 :
FIGURE 6: F1 CHANGES OF ENGLISH L2 Figure 6 demonstrates the decrease and increase of F1 value in closing diphthongs.The decreasing values are shown by the left bars.The right bar indicates the increase of F1 value.It can easily be spotted that there is a right bar for the sound /Tj/ which indicates the significant difference among the diphthongal productions.The F1 value in the sound /Tj/ surprisingly increased instead of decreasing.It simply puts that English L2 tends to widely open the mouth at the end of the diphthong when they pronounced the sound /Tj/.

FIGURE 7 :
FIGURE 7: THE AVERAGE OF L2 FORMANT CHANGES OF EACH ENGLISH CLOSING DIPHTHONG IN BARK SCALE.

TABLE 1 :
THE FORMANT FREQUENCIES AND DURATION OF L2

TABLE 3 STATISTICAL
MEASUREMENT OF DIPHTHONGAL DURATION