YOD-DROPPING VARIATION IN FIJIAN ENGLISH: A SOCIOLINGUISTIC INVESTIGATION

Yod dropping belongs to the field of sociolinguistics that relates to the omission of /j/ sound such as in ‘chew’, ‘tune’, ‘music’, and others. In IPA /j/ sound is similar to /y/ as in ‘yes’. This paper looks at yod dropping for males in Fiji between the ages of 18 to 30 years to evaluate whether or not speakers produce a ‘yod’ before a /u/ vowel. The purpose of this research was to look at (i) if people who speak English in Fiji produce a yod or not, (ii) if the speakers produce a yod, is the production of the yod is conditioned by the phonological environment, and (iii) does the sociolinguistic variable of gender, age, and education impact on the production of yod. This study used quantitative methodology where data was collected using audio recording. The participants read out aloud a paragraph in the English language. This paragraph was transcribed in ELAN software where token words were identified and coded. Jamovi software was used in analysing the data and producing graphical representations of the findings. The major findings revealed that out of the 499 token words, 33.7% did not have a yod. However, 66.3%, which is more than half of the data illustrated that they had yod. The results demonstrate that sociolinguistic variables such as age impact on the production of yod. As in this study, only males were employed; their age group played a distinctive role in influencing the data.


INTRODUCTION
This study looked at yod dropping for males in Fiji between the ages of 18 to 30 years to evaluate whether or not speakers produce a 'yod' before a /u/ vowel. Yod dropping belongs to the field of sociolinguistics that relates to the omission of /j/ sound such as in 'chew', 'tune', 'music' and others. In IPA /j/ sound is similar to /y/ as in 'yes'. Butcher (2020) provides three classes of yod dropping. She explains that in early yod dropping, /j/ is lost after /r/, palatals and initial clusters of a constants + /l/ for example 'rule'[ɹju:l] (Butcher, 2020). Chambers (2002) points that the phonological process that generally involves the simplification of /ju/ to /u/ is referred to as yod dropping.
In addition, Glain (2012) explains that the realisation of /j/ depends on the phonetic environment, on the vocabulary and the variety of English that is being used. According to Chambers (2002) the phoneme /ju/ is the strangest of all English vowels because other tense vowels usually occur with off-glides, only /ju/ occurs with an on-glide. Examples of off-glides include bee /ij/, bay /ej/, bye /aj/, boo /uw/, beau /ow/, boy /oj/ and baugh /aw/ (Chambers, 2002). Simplifying the vowel system by eliminating /ju/ would actually complicate the set of consonant clusters in words such as music, beauty, few, cue and others. Chambers (2002) argues that there are no English words for example /mjij/, /bjej/, /fjow/ or /kjay/ and that yod would not take place in English consonant clusters. Therefore, /ju/ must be a vowel phoneme. This study looked at the following three questions: i) Do people who speak English in Fiji produce a yod or not? ii) If the speakers produce a yod, is the production of the yod conditioned by the phonological environment? iii) Do the sociolinguistic variable of gender, age and education impact on the production of yod?
Finally, there are five sections in this paper, a section on research questions follows after, then the pivotal literature that was reviewed in this study which is followed by the materials and methods section. The results and discussions section is followed by the concluding section that provides a summary of this research.

LITERATURE REVIEW
Studies on yod dropping can be traced as far back as 1880s (Ellis, 1889). In her research, Ellis looked at 'generalised yod drooping' for working class communities based in Norfolk and Suffolk during the 19th century. In another study carried out by Orton et al. between 1962Orton et al. between -1971 illustrated that rural speakers steadily produced yod-less [u:] across C+/ju/ in Suffolk even after 30 years (Butcher, 2020). A similar finding came from Trudgill (1974) which showed that in Norwich all classes exhibited generalised yod dropping. However, in recent studies such as Amos (2007) there seems to be yod retention based on gender to some extent. Interestingly, Amos (2007) found that there was a decline with traditional realisation of older speakers but young females "displayed near-categorical maintenance of yod retention". Britain (2011) used results from their research on the Fenlands and Ipswich to show that the retention rates were higher in these regions. The study found the rendition rate was over 80% for older speakers and 70% for younger speakers.
Yod dropping is also linked to location which was later confirmed by Britain (2014). For instance, generalised yod dropping was prevalent in rural East Anglia but there was a decline in towns and cities (Butcher, 2020). In London and South East accents, the elision of /j/ is common after /t, d, n/ (Britain, 2011). Palatalization such as the approximant /j/ coalesces with the preceding alveolar stop to result in [tʃ, dʒ] is also common in Australian English and General American English (Kazemi, 2015;Wells, 1982). According to Britain (2011) there is an increase in palatalization by the younger generation in both Mersea and Ipswich. Butcher (2020) pointed out that this pattern suggests a correlation between levels of coalescence and closeness to London. However, the data from Britain (2011) study highlighted that young speakers who lived closer to London were more yod-ful. Therefore, Butcher (2020) and Hannisdal (2006) argue that in the case of coalescence geographical location may not be so important after all. Chambers (2002) explains that as a vowel /j/ occurs in minimal pairs with /u/ for example mute and moot, and coo, feud and food, beauty and booty. He points out that the English spelling helps in identifying /ju/ words from /u/ words even though it may not be reliable (Chambers, 2002). Another crucial observation that is highlighted by Chambers (2002) is that the spelling <oo> never represents the phoneme /ju/. The reason for this is that in many English accent it represents the lax vowel /o/ for instance in book, look, foot, and others. Another argument put forward by Chambers (2002) is that the vowel /ju/ is poorly imbedded into English vowel inventory for two reasons. Firstly, the vowel /ju/ is on-gliding diphthongs and secondly its high back rounded nucleus also occupies the same place as the plain vowel /u/ (Chambers, 2002). He further indicates that as /ju/ is poorly integrated, it is more likely to be susceptible to change and dialect variation (Chambers, 2002).
In providing an overview of yod droppings, Chambers (2002) explains that originally, English dialect had two phonemes which are /u/ and /ju/. There has been a tendency to merge the two phonemes by the loss of yod in /ju/ in order to make it distinct from /u/. Wells (1982) highlights that it could have been that the merger originated after palatals where /ju/ lost its yod in a number of English accents. Hughes et al. (2013) clarify that in Scottish, the distinction occurs as RP is fairly conservative with yod-dropping. This sometimes occurs after /l/ such as lute and illusion, and sometimes after /s/, for example in suit (Hughes et al., 2013). Worth noting is that North American English and British English differ with yoddropping. Chambers (2002, p. 6) mentions that British varieties such as Australian, New Zealand and South African English usually retain /ju/ after the non-continuants /t/, /d/ and /n/. However, they often drop yod after the continuants /l/, /s/, /z/ and /θ/ (Chambers, 2002).
In brief, research on yod-dropping can be traced back to 1880s. Studies have shown that retention of yod is also gender specific for example the study by Amos (2007). Also, yod dropping has been linked to location that was confirmed by Britain (2014). Furthermore, Chambers (2002) highlighted that the vowel /ju/ is poorly imbedded into English vowel. Finally, literature indicates that yod-dropping differs between North American English and British English varieties. However, a study on yod-dropping had yet to be conducted in Fiji. Therefore, this study is one of the first to look at yoddropping of participants from Fiji.

MATERIALS AND METHODS
This study used quantitative methodology where data was collected using audio recording. The participants read out aloud a paragraph in the English language. This paragraph was transcribed in ELAN software where token words were identified and coded. The results were validated with statistical analysis. Jamovi software was used in analysing the data and producing graphical representations of the findings. The data of this study was collected using proportionate stratified random sampling or judgement sampling. The participants of this study were 20 males from Fiji between the ages of 18-30. They were tertiary educated individuals. The data collection was carried out with the help of a colleague who teaches at a university in Fiji. He organised the consent and survey forms to be signed and upon consent from the participants I proceeded with data collection. The participants were clearly informed that participation was voluntary, and they could leave the study at any point in time. We also informed the participants that their information was confidential, and no personal data would be misused.

RESULTS AND DISCUSSIONS
This section of the paper presents findings and discussions based on the yod study data. The first question that this study addressed is: Do people who speak English in Fiji produce a yod or not? According to Table 1, the study had a total of 499 dependent variables for 20 participants. Further, the standard deviation for the token words was 0.473. Out of the 499 token words, 33.7% did not have a yod. However, 66.3%, which is more than half of the data illustrated that they had yod. As the 20 participants of the study were only males, it would be interesting to note if the results would have been different if female participants were included. Table 2 looks at the frequency of dependent variable. For the dependent variable frequency, as depicted in Figure 1, a total of 331 yod was produced with for the 499 token words. Figure 1 also highlighted that 168 did not produce any yod for the token words.  Table 3, it can be noted that 25 token words were highlighted by each speaker. Speaker 14 had the maximum number of yod produced with a mean of 0.840 followed by Speakers 11 and 12 with a mean of 0.800. Interestingly, all the three speakers highlighted were males over 30 years of age. Speaker 4 produced least number of yod with mean of 0.320 followed by Speaker 2 with a mean of 0.400. Both of these speakers were males of 18 years. However, more investigation is needed to conclude whether yod is produced more with older males when compared to younger. The second question that this study addressed was: If the speakers produce a yod, is the production of the yod conditioned by the phonological environment? Looking at Figure 2, it can be implied that since a total of 331 yod was produced for the 499 token words which is 66.3% of the total data, yod production is conditioned by the phonological environment. According to Figure 2 less than half the total data of 168 did not produce any yod for the token words. This is 33.7% of the frequency.

Figure 2. Frequency graph for Dependent variables
The final question that this study aimed to address was: Do the sociolinguistic variable of gender, age and education impact on the production of yod? Table 4 represents the data on the gender and age of the participants. Code 1 is for male speakers between 18-30 whereas as code 3 is for male speakers over 31 years. Based on the data, it can be stated that male speakers over 31 years had slightly more yod production when compared to those between 18-30 years. In total, 180 yod were produced with 31 years and above in comparison with 150 for those between 18-30 years. Interesting to note is that 99 token words did not have any yod for male between 18-30 years. To add on, Table 5 illustrates that the p value is lower than 0.05. Since it is 0.004 it can be concluded that the difference between the two groups is significant. This implies that sociolinguistic variable such as age impact on the production of yod. As in this study only males were employed; their age group play a distinctive role in influencing the data. Therefore, it can be claimed that older males produce more yod than younger ones. Finally, in closely examining Figure 3, it highlights that a higher number of the younger group which is 18-30 of age did not produce yod for the token words whereas the older population of 31 years and above did. The ten males over the age of 31 years produced more yod for the token words compared to the younger ones.

Figure 3. Counts of Dependent variables
More investigation is needed to determine if the education background of participants can be a contributing factor for yod production. In the study, the second group of 31 years above produced more yod than the first group of 18-30 years. It can be argued if the younger males had more exposure to different levels of the English language when compared to the older males.

CONCLUSIONS
This study looked at yod dropping for males in Fiji between the ages of 18 to 30 years to evaluate whether or not speakers produce a 'yod' before a /u/ vowel. The aim of the research was to look at (i) if people who speak English in Fiji produce a yod or not, (ii) if the speakers produce a yod, is the production of the yod conditioned by the phonological environment, and (iii) does the sociolinguistic variable of gender, age and education impact on the production of yod. The findings revealed that out of the 499 token words, 33.7% did not have a yod. However, 66.3%, which is more than half of the data illustrated that they had yod. As the 20 participants of the study were only males, it would be interesting to note if the results would have been different if female participants were included. Thus, further study in Fiji is needed in this area for more reliable conclusions. Based on the analysis, it can be implied that since a total of 331 yod was produced for the 499 token words which is 66.3% of the total data, yod production is conditioned by the phonological environment.
The p value was lower than 0.05, it is 0.004. Therefore, it can be concluded that the difference between the two groups is significant. This implies that sociolinguistic variable such as age impact on the production of yod. As in this study only males were employed; their age group play a distinctive role in influencing the data. Therefore, it can be claimed that older males produce more yod than younger ones.