Investigation into the clear ~ dark /l/ allophony in Rhondda Valleys English

1. Abstract

Little research has been carried out in Wales, especially concerning the industrialised Valleys area. Thomas (1984) claims that /l/ is clear in all positions in the southern dialects of Wales and dark in the northern dialects. Walters (2001) then brings forward the most recent findings stating that prevocalic /l/ is clear in all dialects, but the postvocalic /l/ follows a pattern of: back vowels favouring the dark /ɫ/, whilst front vowels inhibit it. Adding to the current research, I have considered the affect of the preceding consonant and the length of the preceding vowels on the dark /ɫ/. The clear ~ dark distinction that is apparent in the younger informants of our research is quite clearly a new development, represented by the lack of dark /ɫ/ in the older informants and the inconsistencies in the results. This being said, the results were as predicted, with certain linguistic environments favouring the dark /ɫ/ over the clear variant. It is then clear to see the similarities between the favourable environments of the dark and vocalised /l/, leading me to suggest that in fact these are both part of the Lenition process. Once the clear ~ dark distinction becomes more apparent, I would expect to see the emergence of vocalisation in the Rhondda Valleys English.

2. Introduction
“The subject of Welsh English is inadequately documented.”

(Thomas, A. 1984:178)

The data for this project was collected by me and Gemma Emslie. We decided to look at the Welsh English dialect, as the linguistic research on this area seemed to be lacking and Rhondda in particular has not been the subject of much investigation. Upon listening to our data and reading the relevant literature, we chose to investigate the quality of rhymal /l/ in Rhondda Valleys English. According to Thomas, /l/ is clear in all varieties of Welsh English. (Trudgill, P. 1984:186). Penhallurick (2004) later claims there is actually a north-south divide for the allophonic variation of /l/, stating that in South Wales /l/ is clear in all positions, and dark in the north. (Penhallurick, R. 2004: 62). This is said to reflect the Welsh language which shows the same variation; clear /l/ in all environments in the south, and dark once again in the north. However, we then found a clear ~ dark /l/ distinction in Rhondda, going against the general consensus that clear /l/ is found in all positions in the Southern Welsh dialect. For this reason, the aim of this discussion is to investigate which environments favour this dark /ɫ/ that has recently become apparent in the area of Rhondda.

3. Welsh English
“[i]f Welsh were not to be used in a significant formal context then it meant, too, that its use in informal contexts would diminish”. (Penhallurick, 2004:98; Cited in Aitchison & Carter 2000:27).
From the 5^th Century onwards, the arrival of invaders such as the Saxons and other Germanic-speaking tribes into Britain, exerted pressure on the Celtic speakers to move into the area now known as Wales. From then onwards, these Celtic speakers would be subject to a long process of Anglicisation – meaning the replacement of Welsh by English as the dominant language. (Walters, J.R 2001:287).
“English is a thoroughly established language of Wales, a language used by and belonging to the welsh people – not that they have sole ownership of it, of course.” (Penhallurick, R. 2004:100).
This anglicisation was accelerated with events that encouraged the rise in Status of the English language, subsequently leading to the demise of the Welsh language. “Under the acts of Union of 1536-1543, English was made the sole language of government and law in Wales.” (Penhallurick, 2004:98). The industrial revolution of the 19^th Century also saw the government teaching working class types ‘a knowledge of the English language’ (Walters, J.R. 2001:287). However, this did not lead to a complete rejection of the Welsh language. Up until the 20^th Century, Welsh still remained the first language in much of rural Wales. It is inevitable though, that its use would only remain in restricted groups of people and continue to decline. The table below shows the decline in Welsh speakers, from 1891 to 1991:

Figure 1. The decline of Welsh Speakers in the 20^Th Century

Walters, J.R (2001:288)

There are noticeable differences between the traditional Welsh dialects of the north and south, especially concerning phonology. These are represented in the pronunciation of the Welsh English dialects that are present today, with a North-West and mid-South divide.

4. The Rhondda Valleys

5. The /l/ Variable

Sproat & Fujimura (1993) claim that the clear and dark allophones are not actually distinct phonological entities, but are simply lighter or darker variants of the /l/ variable. However, for the purpose of this project I will continue to discuss the allophones as the canonical clear or dark variants. The dark, vocalic gesture has an affinity to the nucleus, whilst the lighter, clear /l/ has an affinity to the margin. This reflects my comments regarding the allophones being preferred as either a syllable initial or rhymal segment. The allophones are therefore phonetically distinct in that the clear or light allophone is usually prevocalic and the dark allophone is post-vocalic or a syllabic. (Sproat, R. & Fujimura, O. 1993:292).

Within the rhyme, pre-boundary /l/ varies continuously from the lighter, clear variant to the darker, vocalic variant. The preceding rhyme is responsible for this variation and the /l/ will vary depending on the duration of the rhyme that the /l/ finds itself in. Sproat & Fujimura claim that the /l/ in shorter rhymes is of the clear, lighter variant and the /l/ in longer rhymes will be dark. (Sproat, R. & Fujimura, O. 1993:200)

With regards to the syllable position, consonantal gestures are typically found in the syllable initial position. It is therefore assumed that in the syllable final position this consonantal gesture is weaker and the vocalic gesture takes precedence. The weakening of syllable final consonants is a wide spread phenomenon and considered a natural process. The term Lenition refers to the strength of the segment, which goes through a process of weakening, eventually resulting in a loss of the segment altogether. The strongest sound is the first segment that occurs when the process begins, the weakest sound therefore being the one that occurs just before the sound is lost. This process of lenition is seen as reducing the complexity of the segment. With regards to /l/, the consonantal gesture is lost to make way for a simpler, more vocalic gesture. (Harris, J 1994:120).

The weakening of syllable final /l/ is represented below:
Clear /l/ ~ Dark /l/ ~ Vocalisation

5.2 Vocalisation & Linking

Vocalisation is most likely to occur in the syllable rhyme and only in those dialects which exhibit the clear ~ dark /l/ distinction or those dialects where /l/ is relatively dark in all positions. The dark /ɫ/ is therefore seen as a prerequisite for vocalisation.

Without such conditions, vocalisation is much less likely to occur. Therefore the dialects where /l/ is clear in all positions will not be prone to vocalisation. The emergence of this vocalised dark /ɫ/ is thought to be expected as a natural progression, often being referred to as the emergence of the ‘unmarked’. Unmarked forms tend to be more natural and structurally simpler. (Johnson, W. & Britain, D. 2003:298). Those dialects which have been late to acquire the clear ~ dark distinction are likely to have begun the change to vocalisation much later than those dialects where the distinction is well entrenched.

Following the same pattern as the clear ~ dark /l/ distinction, the preceding vowel can determine whether vocalisation is likely to occur. Preceding long vowels are more likely to be followed by vocalisation than preceding short vowels. Sproat and Fujimura (1993) suggest that the length of the rhyme can affect vocalisation levels. They suggest that a long rhyme will promote an early dorsal gesture, explaining why long vowels are seen to promote vocalisation. Conversely, a short rhyme will tend to inhibit vocalisation. (Sproat, R. & Fujimura, O. 1993:301)

Meuter (2002) supports these findings, summarising that vocalisation is more common after back or central vowels and especially long vowels. (Meuter, A. 2002:50-51). Back vowels, as commented on here by Meuter, were also found to promote the dark /ɫ/ as discussed above.

One complication that arises when distinguishing the two allophones of /l/ is found when considering the properties of the intervocalic /l/. (Sproat, R. & Fujimura, O. 1993:292). The intervocalic /l/ does not always act as a normal post-vocalic variant. Consider the variant when followed by a word or morpheme boundary, e.g. . The /l/ here would not be labelled as ‘prevocalic’ due to the boundary that separates it from the following vowel. However, the /l/ acts as a linking segment to break the hiatus and has a clear realisation. This leads you to believe that speakers retain an underlying /l/, even when it doesn’t occur in the rhyme position and hasn’t been made vocalic. (Johnson, W & Tipton, P. 2007:10). It could be argued here that the /l/ is retained through syllabification into the onset. Consider an area such as the Fens; researched by Dave Britain. With the late emergence of the clear ~ dark distinction, this area shows low levels of vocalisation. Therefore, when the /l/ occurs before a vowel, vocalisation is almost non-existent. Instead of vocalising, they retain the clear linking /l/ in these pre-vocalic environments. Vocalisation then, seems to favour preceding a consonant, rather than a vowel; as shown by the linking /l/ that occurs pre-vocalically where /l/ would have historically been weakened. It could therefore be suggested that the status of /l/ in these pre-vocalic environments represents how established vocalisation has become within that dialect. (Johnson, W. & Britain, D. 2003:310).

~ [fu:ɫ], because people are conscious of their speech; this consciousness being modelled by RP, not local norms. (Trudgill, P. 1984:186)
J.C Wells goes on to acknowledge a clear ~ dark allophony in the anglicised areas of South Wales. However, he does not include Rhondda in this generalisation, commenting that “in other parts of the south, e.g. Merthyr, Rhondda, Neath, /l/ tends to be relatively clear in all positions.” (Wells, J.C 1982:390).
Although the majority of the literature on Welsh English accent features claimed that /l/ was clear in all positions in South Wales, upon analysing our data we noticed that the younger informants of Rhondda did in fact display a clear ~ dark /l/ distinction. Clear /l/ was however present in all contexts for the older informants. This interesting finding led us to investigate this clear ~ dark /l/ allophony for our project.
6. Subjects & Data Collection
As previously mentioned, the research that has been carried out on this area was sparse. This in mind, we went to Rhondda not having decided on a feature. Data was collected from two different age groups; 18-25 and 55 and over, speaking to both Males and Females. At this point, we were not sure on our variable so could not rule out any socio-variables that may affect our data. We were therefore free to listen to the data, noting any interesting features that occurred before deciding on our chosen variable. Before arriving at Rhondda we contacted employees of Rhydfelin College to locate our younger informants. These contacts then pointed us in the direction of a town called Maerdy to find our older informants, where we located an archive group and a community centre. Both Rhydfelin and Maerdy are located within the Rhondda district. To collect the most natural and informal data possible we approached all of our informants in an environment which was comfortable to them, trying to engage them in casual conversation. Although the informants were made aware of our research purposes, they were not informed of any potential variables we may have been listening for. Contrary to the claim made by Thomas (see above, 5.3), /l/ is not a variable that I believe would be changed by a conscious, socially aware speaker. Therefore, I think it safe to assume that the noted /l/ variables were taken from complete informal speech. The data collection was far easier with the older informants, especially when enquiring about the history of their town. It was not so easy to find younger people to speak comfortably for any length of time. We found this to be especially true of the younger male informants and finding an informal environment quiet enough to record their speech proved to be very difficult. Our findings are based on the speech of 8 informants, all of whom have lived in the Rhondda area for most of their lives.
7. Results & Constraints
After having collected the data from Rhondda, Gemma and I thought it was best to first analyse the data individually. Once coming to our own conclusions, we came together to discuss our results. Any differing results were listened to again, and a conclusion was reached together for the tokens we were unsure of. As mentioned before by Sproat & Fujimura (1993), clear and dark /l/ are not two complete distinct entities and in fact range from the lighter variants, to the darker variants. However, for ease of analysing our data, we decided to label our tokens as either clear or dark. We were careful to only label the dark tokens as so, if we were sure that they were completely dark. We selected 33 tokens from each of the age groups, making sure to cover all the phonetic environments for the allophony of rhymal /l/. These were then put into tables along with the phonetic realisations of /l/ and its preceding or following environments. After creating these tables for both the younger and older age groups, it soon became clear that dark tokens were only found in younger informants. For the purpose of the project, I will only be concerned with the environment in which the dark /ɫ/ occurs in these tokens of the younger informants.

The following table shows the percentage of clear and dark /l/ in the two age groups:

Figure 3. The percentage of Clear & Dark /l/ in the two age groups.

Number of tokens:

Age Group	18-25	55+
Dark /ɫ/	9	0
Clear /l/	24	33

Meuter (2002) states that vocalisation is a sound change that is put into progress by younger people. (Meuter, A. 2002:22). This seems to also be the case with dark /ɫ/, which you would expect to find as dark /ɫ/ is a prerequisite for vocalisation.

/l/ has been labelled as a segment that is generally not affected by socio-variables, however I agree with Meuter in that age is a socio-variable to be taken into account here; as proven by our data.
7.1 The effect of preceding consonant on Syllabic /l/

Previous studies have shown that preceding coronals favour the dark /ɫ/ the least and a preceding dorsal or labial is most likely to favour the dark /ɫ/. (Johnson, W & Tipton, P 2007:9). Consider the Fens data with regards to vocalisation. It is reported that labials and dorsals promote vocalisation over a preceding coronal. These findings are not universal, but tend to be the same for most English dialects that show the clear ~ dark dichotomy. One point to note here is that with my data and that from the Fens, a preceding coronal is never the most favourable preceding consonant for vocalisation or dark /ɫ/. Referring back to Sproat & Fujimura’s claim that /l/ is a complex segment with two gestures; the consonantal coronal gesture and the vocalic dorsal gesture. They suggest that the articulation of the clear /l/ consists of the coronal gesture preceding the dorsal gesture. The reverse is applied for the dark /ɫ/; the dorsal gesture preceding the coronal gesture. Therefore, due to assimilation of place of articulation, we would expect to find that dorsals promote the dark /ɫ/ and the coronals promote the clear /l/.

Figure 4. The effect of preceding consonant on syllabic /l/.

Number of tokens:

Preceding Consonant	Coronal	Labial	Dorsal
Dark /ɫ/	0	1	1
Clear /l/	2	4	2

As expected, the dorsal and labial preceding consonants do favour the dark /ɫ/ much more so than the preceding coronal gesture, thus the preceding coronal being the favourable environment for the clear /l/. These results are not surprising when considering the articulation of the two allophones. As mentioned, the clear /l/ consists of a more prominent consonantal coronal gesture and the dark /ɫ/ consisting of a more prominent vocalic dorsal gesture. It is due to this prominent vocalic gesture that the dark /ɫ/ and vocalised /l/ are of similar quality and seem to favour the same preceding environments. Although there were equal tokens of dark /ɫ/ following a labial and a dorsal, the percentage was higher for the dorsals as the dark /ɫ/ occurred in one instance out of two. We did notice however, that the dark /ɫ/ that followed the labial was almost vocalised . This /l/ in question was preceded and followed by a labial and seeing as the articulation of the vocalised /l/ ~ transcribed by some as /o/ or /ω/~ is also labial, it is not surprising to see that the labials perhaps promote vocalisation. Meuter (2002) adds that “a preceding labial favours vocalisation whereas coronals and dorsals disfavour vocalisation” (Cited in Spero 1996: 56, 57, 62). This is one area in which the dark /ɫ/ and the vocalised /l/ are not exactly the same. It seems that whilst dorsals promote the dark /ɫ/, they particularly disfavour the vocalised /l/. (Meuter, A. 2002:25, cited in Spero 1996:56, 57 & 62).

Although the findings here seem to be as expected, there were still cases of dorsals and labials favouring the clear /l/. Notice here that prevocalically, after any consonant, there are no instances of dark /ɫ/. In the environment of Rhymal /l/ followed by a vowel, the clear /l/ is retained and realised in the following onset. Consider the examples
, , and . provides a preceding coronal for /l/, therefore expecting to find a clear /l/ here. However, especially with the example of , the preceding dorsal should favour a dark /ɫ/. This dorsal token exhibiting a clear /l/ is followed by a vowel, therefore explained by the ‘Linking’ phenomena explained above (5.2).

As discussed, when followed by a vowel the /l/ variant is clear. When /l/ is followed by a consonant or a boundary however, the variant is dark. Meuter (2002) adds to these findings, concluding that “both age groups vocalise /l/ the most when there is no segment following….vocalisation is, furthermore, favoured by both age groups in the environment of a following /w/.” (Meuter, A. 2002:51-52). Interestingly so, the only consonant following a dark /l/ was /w/ in .

7.2 The effect of preceding vowel on /l/
(i) When considering the effect of the preceding vowel on /l/, firstly it is important to look at the position of the vowel; [+/- Back].

J.R.Walters claims that the prevocalic /l/ is always clear, but the postvocalic /l/ generally follows a patter of:

• 
  Clear /l/ generally occurs after a front vowel, and
  
• 
  Neutral or dark /ɫ/ after a back vowel.
  

(Walters, J.R. 2001:294)

Walters does not however make any statements about central vowels, vowel length or diphthongs. Seeing as there are some occurrences of a dark /ɫ/ after a central vowel, I thought this was important to consider.

Figure 5. The effect of preceding vowel position on /l/.

Number of tokens:

Preceding Vowel	Front	Back	Central
Dark /ɫ/	2	4	1
Clear /l/	11	4	3

As our results show, front vowels did in fact favour the clear /l/ and there were more occurrences of dark /ɫ/ following a back vowel than a front vowel. These findings concur with the predictions made and can be explained by assimilation of place of articulation as with the preceding consonants. Clear /l/s consist of a stronger coronal gesture towards the front of the oral cavity and the dark /ɫ/ involves a stronger retraction towards the back of the mouth. You would therefore presume that the /l/ would be favoured by a preceding front vowel and the back vowels would promote the dark /ɫ/ that has a stronger vocalic gesture towards the back of the oral cavity. The central vowels seemed to favour a moderate amount of dark /ɫ/s, less so than the back vowels, but more so than the front vowels. This result fits in perfectly with the analysis of the preceding vowels favouring clear or dark /l/ based on their place of articulation. The vowel order for favouring the dark /ɫ/ is displayed below:

Back Vowel > Central Vowel > Front Vowel
There were however cases of front vowels favouring the dark /ɫ/s and back vowels favouring the clear /l/. When first looking at these anomalies, I thought perhaps this clear /l/ following the back vowel could infact be linking. However, all of the environments concerning these clear /l/s that follow a back vowel are either pre-consonantal or at the word boundary. As we know, linking /l/ occurs pre-vocalically and there were no V_V sequences to suggest this may be the case. These unusual cases can be due to the fact that the clear ~ dark distinction is still relatively new in this dialect and the results would perhaps be more consistent in future data collections.
(ii) The second consideration of the preceding vowel is the length. As shown above, Walters makes a generalisation regarding the position of the preceding vowels, but makes no reference to the vowel length. Sproat & Fujimura claim that:
“the quality of the pre-boundary /l/s varies continuously with the pre-durations of the rhyme in which the post-vocalic /l/ finds itself in.”

(Sproat, R. & Fujimura, O. 1993:300)

The generalisation is then made that the /l/ in shorter rhymes is lighter and of the clear kind, and the /l/ in longer rhymes is usually dark.

Sproat & Fujimura go on to explain that the long vowels promote the early and longer dorsal gesture of the dark /ɫ/, whilst the shorter vowels inhibit it. The more prominent the dorsal gesture, the more likely the coronal one will fail. (Johnson, W & Britain, D. 2003:302).

Figure 6. The effect of preceding vowel length on /l/.

Number of tokens:

Preceding Vowel	Long	Short	Diphthong
Dark /ɫ/	3	2	2
Clear /l/	3	13	0

Once again our results do concur with our predictions, seeing that shorter vowels do indeed favour a lighter-clear /l/ and the longer vowels favour the dark variant. Although there are more occurrences of dark /ɫ/ following a long vowel than there are following a short vowel, there are still equal amounts of the clear and dark variant following the long vowel. Likewise with the other findings, I seem to have the patterns that suggest this language change is moving in the right direction (towards the unmarked vocalisation), but the results are still fairly sporadic. The diphthong produces a nice clear pattern however, favouring the dark variant 100%. The diphthong is essentially a long vowel, so based on the above hypothesis; the dark /ɫ/ was expected here. In Meuter’s vocalisation study, she found that in both of her age groups, vocalisation was most likely to occur after a diphthong. (Meuter, A. 2002:49). An interesting comparison to make here is that the lexical item exhibited the clear allophone in the 55+ age group, but the 18-25 year olds displayed the dark variant. This is a perfect example of how the age groups have differed in their development of the clear ~ dark /l/ distinction.
To summarise my findings, I would say that the results have generally mirrored our predictions. I found that the preceding consonants that favour the dark /ɫ/ are dorsals, followed by labials and then coronals. It was also clear that back and long vowels (including diphthongs) do favour the dark /ɫ/, with front vowels promoting the clear variant. This being said, the results were not completely conclusive and there were some anomalies in the data. The first reason for this being that once we had established that the older informants used clear /l/ in all positions; we were then left with only half the data to use when investigating the environments that favour the dark /ɫ/. Only 9 tokens of the dark /ɫ/ were then available to analyse. The second and main reason for the anomalies being that the clear ~ dark dichotomy that is present in the Rhondda Valleys English today is considered to be a very new phenomenon. Once this distinction has become more entrenched, we can perhaps hope to find more consistent and reliable data. Interestingly though, even with so little tokens to analyse, we have still found the expected patterns in our data. Our findings certainly pave the way for further investigation. Perhaps finding more of the younger informants and V_V sequences to analyse would improve our findings some what.

8. Conclusion

The Welsh English spoken in the south of Wales has been described as having clear /l/ in all positions. The only exception of Cardiff was made, claiming that it represents its anglicised features by having the clear ~ dark distinction found in many dialects of English today. J.C.Wells went so far as to classify Rhondda in the group of dialects showing ‘clear /l/ in all environments’.

However, along with those of Walters (2001), our findings suggested that the clear ~ dark distinction is beginning to develop in the Rhondda Valleys. This distinction is a new development, represented by the dark variants only being present in the speech of the younger informants and the older informants showing clear /l/ in all positions. Although exhibiting this distinction, the results of the younger informants were less than conclusive, with only 9 tokens to analyse and some unexplainable anomalies. Nevertheless, these results confirmed the predictions put forward based on general literature of favouring environments for the dark /ɫ/ and vocalisation. It was interesting to see that the favoured linguistic environments for the dark /ɫ/ were very similar to that of vocalisation. Seeing as dark /ɫ/ is a prerequisite for vocalisation, this would suggest that they are both stages in the process of final-weakening – Lenition.

With further research I would expect to see the clear ~ dark distinction more stabilised in its phonetic environments, occurring more often in the innovative younger informants. In time to come, this distinction will hopefully lead to the emergence of /l/ vocalisation in Rhondda Valleys English.

Word Count: 5422
9. Appendix

18-25 Age Group ~

Lexical Item	Realisation of /l/	Location	Phonological environment
unbelievable	/l/	C_#	labial_#
people (on)	/l/	C_V	labial_ɒ
people	/l/	C_#	labial_#
apple (whatever)	/ɫ/	C_C	labial_labial
couple (of)	/l/	C_V	labial_ə
normal	/ɫ/	V_#	ə_#
Google	/ɫ/	C_#	dorsal_#
local (and)	/l/	C_V	dorsal_ə
locals	/l/	C_C	dorsal_coronal
middle (of)	/l/	C_V	coronal_ə
metal	/l/	C_#	coronal_#
(two) mile	/ɫ/	V_#	ʌɪ_#
(a) mile	/ɫ/	V_#	ʌɪ_#
girl (once)	/l/	V_C	ɜ:_labial
old	/l/	V_C	ɒ_coronal
sold	/l/	V_C	ɒ_coronal
bowled	/ɫ/	V_C	ɒ_coronal
film	/l/	V_C	ɪ_labial
help	/l/	V_C	ɛ_labial
(key) skills	/l/	V_C	ɪ_coronal
(chav) central	/l/	V_#	ə_#
school	/l/	V_#	u:_#
wheelchair	/l/	V_C	i:_coronal
already	/ɫ/	V_C	ɔ:_coronal
all (the)	/ɫ/	V_C	ɔ:_coronal
all (day)	/ɫ/	V_C	ɔ:_coronal
hotel (via)	/l/	V_C	ɛ_labial
hotel	/l/	V_#	ɛ_#
hell	/l/	V_#	ɛ_#
well (I)	/l/	V_V	ɛ_
aswell	/l/	V_#	ɛ_#
full (timers)	/l/	V_C	ʊ_coronal
hill	/l/	V_#	ɪ_#

55+ Age Group ~

Lexical Item	Realisation of /l/	Location	Phonological environment
unbelievable	/l/	C_#	labial_#
people (who)	/l/	C_V	labial_u:
couple (of)	/l/	C_V	labial_ə
trouble	/l/	C_#	labial_#
football	/l/	V_#	ɔ:_#
electrical (fault)	/l/	C_C	dorsal_labial
(a) bicycle	/l/	C_#	dorsal_#
(the) bicycles	/l/	C_C	dorsal_coronal
cycle (of unemployment)	/l/	C_V	dorsal_ə
kettle	/l/	C_#	coronal_#
hospital (at)	/l/	C_V	coronal_ə
miles	/l/	V_C	ʌɪ_coronal
compile (archives)	/l/	V_V	ʌɪ_ɑ:
Ferndale	/l/	V_#	ei_#
railway	/l/	V_C	ei_labial
girls	/l/	V_C	ɜ:_coronal
coal	/l/	V_#	ɒ_#
coal (pit)	/l/	V_C	ɒ_labial
old	/l/	V_C	ɒ_coronal
milk	/l/	V_C	ɪ_dorsal
difficult	/l/	V_C	ʊ_coronal
skills (centre)	/l/	V_C	ɪ_coronal
health	/l/	V_C	ɛ_coronal
building	/l/	V_C	ɪ_coronal
central (heating)	/l/	V_V	ə_i:
school	/l/	V_#	u:_#
schools	/l/	V_C	u:_coronal
always	/l/	V_C	ɔ:_dorsal
well (I)	/l/	V_V	ɛ_ ʌɪ
aswell	/l/	V_#	ɛ_#
well (yeah)	/l/	V_C	ɛ_coronal
full name	/l/	V_C	ʊ_coronal
hills	/l/	V_C	ɪ_coronal

10. References:

Harris, J. (1994). English Sound Structure. Oxford: Blackwell.
Johnson, W & Britain, D. (2003). Appears in revised form as L vocalisation as a natural phenomenon: Explorations in Sociophonology in Language Sciences (2007)
Johnson, W & Tipton, P. (2007). Language Variation and English Phonology. Class notes; University Of Essex. Colchester: Essex.
Meuter, A. (2002). L Vocalisation and labiodental variants of /r/ in the speech of Colchester primary school children – the acquisition of sound change? MA Dissertation. Colchester: Essex University.
Penhallurick, R. (2007), ‘English in Wales’, in D. Britain (eds) Language in the British Isles. Cambridge: CUP
Penhallurick, R. (2004), ‘Welsh English: phonology’, in E.W. Schneider. A Handbook of Varieties of English: Phonology. Mouton de Gruyter: Berlin.
Sproat, R. & Fujimura, O. (1993). Allophonic variation in English /l/ and its implications for phonetic implementation. Journal of Phonetics 21.
Thomas, A. (1984) ‘Welsh English’, in Trugill, P. (1984) Language in the British Isles, Cambridge: CUP
Walters, R. J. (2001), ‘English in Wales and a ‘Welsh Valleys’ accent’, in World Englishes, Vol. 20. No.3. Blackwell Publishers: Oxford
Wells, J.C. (1982) Accents of English, Vol. 2, Cambridge: CUP

LG405 ~ Page Of