WELL, WELL, WELL: VARIATION IN DRESS VOWEL REALISATIONS BEFORE LATERAL /L/ IN AUSTRALIAN ENGLISH

Corresponding Author: Ounu Zakiy Sukaton E-mail: ounu.zakiy@machung.ac.id ARTICLE INFO ABSTRACT Received: 20-07-2020 Accepted: 11-10-2020 Published: 12-10-2020 Volume: 4 Issue: 2 DOI: https://doi.org/10.33019/lire.v 3i2.90 The ongoing el-æl merger in Australian English has been informally recognized by Australians especially those who have experience of contact with Victorians. This study aims to investigate the correlation of speakers’ sex and origin with their /el/ production and how speech styles influence their production. Two male speakers of Australian English from Victoria and South Australia were recorded while reading texts, doing interviews, and having casual conversations. The recordings were then transcribed and analyzed by using various software to describe their /el/ productions. The result of this study is both male subjects are able to produce considerable variations in their /el/ productions. The production of the Victorian male speaker confirmed the findings of previous studies while the SA male speaker showed variations of /el/ similar to back vowels. Speech styles do not significantly affect the variations of /el/ production. Conclusion: The ongoing merger of el-æl in Australian English might be spreading from Victoria through diffusion to its neighboring states. However, more studies should be conducted in order to confirm this suspicion. Other suggestions include customized reading passages and better semi-structured interviews KEYWORDS


INTRODUCTION
A procedure known as 'merger' is where a phonemic contrast is collapsed because of sound change (Hay, Warren, & Drager, 2006). Thus, two phonemes which were already distinct in a language variety are being acknowledged as a single phoneme. This occurence is normal and has been widely observed by sociolinguists. One of the instances of a merger is the merger of NEAR (/iə/) and SQUARE (/eə/) in New Zealand English (Gordon & Maclagan, 2001;Hay et al., 2006). Gordon and Maclagan (2001) did a thorough investigation of Christchurch school students whose age were 14 to 15 since 1983. Their investigation proposed that most students produced clear distinction of NEAR (/iə/) and SQUARE (/eə/) vowels in the recording made in 1983 albeit a few varieties were available and there was no clear tendency of converging toward either vowels. In contrast, they found a practically complete merger toward NEAR (/iə/) in 1998. One thing that Gordon and Maclagan didn't consider is the manner by which the students' creation might be different in reading and casual discussion settings. On the off chance that they had tests from students' creation in casual discussions, they may have the chance to recogn ize the merger sooner. Hay et al (2006) examined the results of this merger in the context of how people will perceive speech. They gathered 71 subjects from University of Canterbury to do perception related tasks by listening to recordings performed by 2 male and 2 female speakers of New Zealand English. The

METHODOLOGY
In the paper, two recordings of one Victoria male speaker by the pseudonym of Ferb and one South Australia male speaker by the pseudonym of Ham will be used. Both speakers are similar in age and level of education. Prior to moving to Canberra for education, they both spent much of their time growing up in their respective hometowns. Ham, however, speaks a second language and he has lived in China for 3 months while Ferb is a monolingual in Australian English. The recordings were done in a secluded space to eliminate noise. ZOOM H5 recorder and head mounted microphone were utilised in all of the recording processes. Three kinds of recording were done by all of the participants. First, three passages which are the North Wind and the Sun, the Grandfather, and the Rainbow passage were read by the participants and recorded. Second, the participants were asked to do a recorded interview for 20 to 30 minutes. Finally, the participants documented themselves having 20 to 30 minutes discussion with their friends or family members. The aim of providing various styles of recording is to elicit speech output from the participants in reading, formal, and casual context. The recordings will subsequently be anonymised using Audacity (AudacityTeam, 2017) and ELAN (Lausberg & Sloetjes, 2009) was used to transcribe the recordings. ELAN orthographic transcription must then be forwarded to PRAAT (Boersma & Weenink, 2018). In this paper, all measurement of the formants of vowels will be done in PRAAT. The next step is to calculate the average value of the different vowels that each speaker generates. The first and second formant of Australian English vowels in Cox and Palethorpe's (2007) illustration will be counted four times in each recording for a total of 12 tokens per vowel (see Di Paolo, Yaeger-Dror, & Wassink, 2011 for reference number of tokens). As per Di Paolo et al. (2011), the conditions from which the measurements of vowels are taken should be in a neutral phonetic consonant setting. If such environment is not provided in the data, however, some convenient samples will be selected instead and some notes followed. In short, two vowel plots (one for each speaker) will be presented in the study containing the mean value of each vowel.
Afterwards, Lobanov (1971) normalisation method will be applied to all of the vowel plots. This approach is chosen because variations in formant values caused by physiological factors can be factored out. Additionally, when all the vowels from the speakers' inventory are included, it

RESULTS
Ferb's and Ham's vowels have different concentration points in their vowel plots when the plots are normalised ( figure 1 and figure 2). Compared to Ham's, Ferb's vowels are concentrated more to the front. On the other hand, Ham tends to produce vowels to the bac k of his tongue. If we have a look at figure 1 and 2, we can see Ferb's and Ham's variability in producing vowels. Ferb and Ham have managed to produce distinct/e/ and /ae/ in most environments th roughout the recordings. We can see it from their/e/ and /ae/ measures (see table 1). Ferb's/e/ and/ae/ output is more stable than Ham's is. We can see from table 1 that Ham has more variations in his / her / its performance (represented by the standard f1 and f2 deviation values). In fact, the value of Ferb's f1 in /e/ is lower than that of Ham which means that Ferb's /e/ is very high relative to Ham's. All the vowel measurements were performed at 50 percent of the length of the vowels since this paper is only interested in monophthong (see Di Paolo et al., 2011 for reference of the vowel measurement).  The results of this study is more or less in line with what Cox and Palethorpe (2004) as well as Loakes et al. (2017) found in their study. Ferb, who is from Melbourne greater area, has a tendency to produce a variety of /el/ with similar quality of /ael/ as illustrated in Figure 1. What was different is that Ferb also produced /el/ with the quality of back vowels during self -recording as illustrated in Table 2. This result was not found in those two previous studies. Moreover, the /el/ production with the quality of back vowels is dominant in Ham's data. According to these specific results, further discussion will be provided in the next section.

DISCUSSION
It's clear that Ferb and Ham have different space arrangement for their vowels. When we go back to table 1, we 'd assume that, due to the standard deviation values of his f1 and f2, Ham would most likely have more differences in his DRESS vowel realisations. However, we can observe in/wVl/ environment that Ferb has more variations in the realisation of his DRESS vowel, particularly in the range of his f2 (see figure 3). Figure 3 indicates Ferb develops four DRESS vowels with qualities of f1 and f 2 close to front vowels. That finding is in line with what was found in the analysis of Cox and Palethorpe (2004) and Loakes et al. (2017). Their Victorians subjects produced similar vowels as in figure 3 to what Ferb displays. What the previous study did not reveal, however, is the fact that Ferb also produces three DRESS vowels with similar qualities of f1 and f2 to back vowels and 1 DRESS vowel with similar qualities of f1 and f2 to central vowels. The latter findings are surprising since, so far, no research has found that /el/ realisations can produce vowels with back vowel formant qualities.
The realisations of DRESS vowel in /wVl/ environment do not seem to be affected by speech style in this study. But for both Ferb and Ham, there's only one token of 'well' in reading context. On the one hand, this particular problem does not seem to influence the assessment of Ham's DRESS vowel realizations in /wVl/ setting since none of his realizations in other environments have similar f1 and f2 consistency to his DRESS vowel. Ferb's 'ok' token in reading, on the other hand, is the only realization of DRESS vowel in /wVl/ setting that has comparable consistency to its central vowels. Ferb's 'well' token in reading, on the other hand, is the only realization of DRESS vowel in /wVl/ setting that has comparable consistency to its central vowels. Therefore, to decide the role of speech style in DRESS vowel realisations in /wVl/ setting, more 'well' tokens in reading activitywould be a great addition.
It is unusual to be able to observe some variations of DRESS vowel realisations in Ham's results, although it is not an uncommon occurrence as well. Labov (2007) proposed that languages could change through transmission and diffusion. Language change through transmission occurred in the same speech culture across different generations. The younger generations adopted what the older generations produced, and further advanced the characteristics that distinguished them from what the older generations produced. By comparison, language change by diffusion is the result of the cultures of speech being in contact. The change is progressing slower in this respect and is less structured than transmission. Clark and Watson (2016) researched the language shift argument of Labov by studying different realisations of lenited / t / in Liverpool and its surrounding cities, Skelmersdale and St. Helens. Their study results backed Labov 's claim to language change through transmission and diffusion. The Liverpool variant of /t/ are /tθ,̳ θ,̳ h/ and these variants are largely retained by the town of Skelmersdale. Clark and Watson suspected that it happened because of massive migration of people from Liverpool to Skelmersdale in the 1960s and ongoing contact between two speech communities (p. 59). Most of the younger generations of Skelmersdale demonstrated the use of /tθ,̳ θ,̳ h/ in their daily speech rather than their peers in St. Helens. In St. Helens, the use of /tθ,̳ θ,̳ h/ variants are slowly being replaced by /ʔ/. /ʔ/ has been observed longer in Manchester rather than

CONCLUSION
This study discusses the probability of differences in the Australian English prelateral DRESS vowel. While this research does not explicitly compare DRESS vowel and TRAP vowel prelaterals, the realizations of DRESS vowel prelaterals in this research might provide further insight into the ongoing merger of el and ael in Australian English. Variations in DRESS vowel realisations in /wVl/ context can be found from male subjects that were not previously examined. In addition, variations in DRESS vowel realizations in /wVl/ environment are surprisingly noticeable from SA's male subject. Although this finding is surprising, this study attempted to clarify how this phenomenon could have occurred in limited detail. Additionally, both subjects produced realisations of the prelateral DRESS vowel with the formant qualities of back vowels. This result indicates that more things are happening in this ongoing merger than simply mergering el-ael. Finally, there would seem to be no important association between the frequency of prelateral DRESS vowel differences associated with speech styles. However, there are some other things to consider. First, there are not enough tokens in the data to fully support this claim, particularly in reading background recordings. Second, more tokens will be required to perform a statistical test There are some suggestions for potential studies in Australian English about the ongoing elael merger. First, tracing how far the phenomenon has spread from its original source would be useful. The fact that SA speaker is demonstrating variations of prelateral DRESS vowel may mean that el-ael merger is not limited to people from the Victoria state. Second, it would also be useful to further explore the connection between speech style and el-ael merger. A personalized reading passage (e.g. a reading passage about 'celery' and 'salary') needs to be made to get the required amount of el or ael tokens. In addition, a better semi-structured interview and self-recording (e.g. talk about 'well,' 'celery,' or 'salary') needs to be done. Furthermore, more tokens can be created by increasing the number of people in future studies. Ultimately, a detailed review is also suggested on why there are certain prelateral DRESS vowel realisations with the formant q ualities of back vowels.