page 1 of 28 Abstract My project focuses on consonant cluster simplification (the variable of -t/-d deletion) and the variation of this feature in Britain. This paper looked at speakers from two areas of Britain, Bridlington and Colchester to see if there was any variation between the two dialects. This paper is hoping to find evidence that there is variation between the two dialects and that this feature is changing in some way but the what this paper will probably find, as has been found in all other dialects of English, that this variable deletion is quite predictable in a general way. It may however allow for some important differences of social, dialectal and individual patterning. The purpose of the paper is to hypothesize a ranking order to see if my findings correlate with previous studies. The findings from my data actually proved that the hypothesis was correct to a certain extent, in that the strongest determiner in influencing -t/-d deletion was the effect of the following segment. This paper also found some interesting data regarding the use of deletion of regular past tense endings. page 2 of 28 Introduction A lot of sociophonological work (although most of it sociolinguistic) has already been carried out on the well-known process of -t/-d deletion in consonant clusters by English speakers. Almost all of the research carried out on this topic has been done on North American, African American Vernacular English (AAVE) and New Zealand dialects of English. This study will compare the consonant cluster reduction in two dialectal areas of Britain, Bridlington in Yorkshire and Colchester in Essex This variable has been extensively studied in all varieties of English and has been studied by linguists and sociolinguists such as Labov, Wolfram, Patrick and Guy to name just a few. Published studies, so far, have found that the deletion of -t/-d has been favoured in a number of ways which include the sonority of the following environment, the presence of the two preceding consonants and the absence of stress among them. Native speakers of English have been found more likely to delete a final -t/-d when it is part of a word stem (e.g. She walked past) rather than when it forms the past tense ending (e.g. She passed the exam.) Speakers of all English dialects are also more likely to delete when the following segment begins with a vowel than when a consonant follows. To carry out this study, I will choose a random sample of seven speakers in total, ranging in ages and I will compare the two areas in terms of preceding environment, following environment and grammatical category. I will compare each speaker to other speakers from their area and with other speakers of the same age group. I intend to study a variety of speakers to get an accurate view of how consonant cluster simplification occurs in each area and to be able to discuss both the phonological and sociolinguistic factors affecting -t/-d deletion. The aim of this project is to see if -t/-d deletion occurs in the same way as previous studies carried out on North American and New Zealand dialects of English have found. I also aim to see if there is any variation within British dialects of English and in particular if there are any patterns emerging in either dialect. I will use published studies to compare my findings and to draw accurate conclusions from my results. page 3 of 28 Maps showing the Location of Bridlington (www.multimap.com) Maps showing the Location of Colchester (www.multimap.com) page 4 of 28 Theoretical Issue The theoretical issue in my project is to see if the same constraints on t/d deletion appear to present themselves in British dialects as they do in North American and New Zealand dialects of English. Part of this issue will also be concerned as to whether there is any variation within Britain itself from the results of my data. To determine whether my results reflect the same constraints as those found in North American and New Zealand studies, I will first need to construct a literature review of previous studies on this topic. Many of the studies previously carried out on the variation of t/d deletion have been primarily concerned with the social constraints on this variable rather than the phonological constraints. Sociolinguists such as Labov, Wolfram and Shuy have all carried out sociolinguistic studies of this variable. One of the first studies carried out on the internal linguistic constraints was by Labov, Cohen and Robins in 1965. In this paper they looked at the speech of black speakers in South Harlem in America. Other studies that followed have come from Wolfram (1969), Fasold (1972) and Labov (1972). All these papers found that in all English dialects that were studied, they produced the same important factors that favour t/d deletion: The sonority of the following environment, The presence of the two preceding consonants, By absence of stress By homogeneity of voicing in the cluster and; By the absence of any grammatical function of the deleted segment. As we have seen this variable appears to operate in the same way in all dialects of English in all the studies that have been carried out to date. The notation below (as presented by Chomsky and Halle 1968) shows that t/d is variably deleted after a consonant at the end of a word and that this occurs more in unstressed syllables such as biggest. It also variably deleted when a third consonant precedes as in next or when there is no preceding grammatical boundary. t,d 0 / - str > [+ cons] < 0 > _ + cons - voc The angled brackets that follow the dash indicate that the first segment of the following word favours deletion if it is [+consonant feature] and the [- vocalic feature]. This generates the declining simplification by the manner of the following segment as obstruent > liquid > glide >vowel. Some different theories have been proposed as an alternative account and to go beyond the regularity of the basic phenomenon. An alternative account of t/d deletion was proposed by those in favour of Optimality Theory in the mid nineties, in which the ordering of a set of universal constraints produced an output dictated by the minimal violations of a set of universal constraints, in a dialect specific ordering. Reynolds (1994) said that to produce a variable output, the theory might be modified to allow variable ordering. Kiparsky (1994) proposed that the ‘exponential relationship could be accounted for by an exploded optimality theory constraint that would have the effect of processing clusters once, twice or three times. Past Tense Marking (a) He washed the car. He seemed funny (b) He washed everyone’s car. He seemed odd. page 5 of 28 In the examples above the /t/ and /d/ are the realisation of the past tense ending –ed, whereas in previous examples such as most, best etc they do not have this function. In all dialects of English we could therefore say that t/d deletion is less likely to occur in contexts where t/d marks the past tense. Evidence from previous studies has shown that this is true as the t/d are the only indication of the tense of the verb. The Linguistic constraints (whether it is followed by a consonant/vowel and whether or not it encodes past tense marking) predict that deletion is more likely to occur in the first set than in the second set. Language Variety Standard American English White Working Class American Eng Black Working Class American Eng Puerto Rican Working Class English Percentage deletion in non – ed clusters (followed by a consonant) 66 Percentage deletion in –ed clusters (followed by a consonant) 36 Percentage deletion in non ed clusters (followed by a vowel) 12 Percentage deletion in –ed clusters (followed by a vowel) 3 67 23 19 3 97 76 72 34 93 78 63 23 This table (Radford et al 1999: 59) shows evidence from a number of dialects and shows that Puerto Rican speakers tend to delete t, d more frequently in than speakers of American English. This table also provides evidence that social factors as well as linguistic factors play a part in the variation of this feature. William Labov says that language is characterised by ‘structured heterogeneity’, in that a set of social, interactional and linguistic factors have complex effects on the linguistic forms found within a speech community. (Radford et al 1999: 60) I chose to study the phonological variable t/d because there has been little work carried out this feature in Britain. Although there have been very similar findings in all other dialects of English, I hope to find something new in my data and in particular some variation between my two British dialects. page 6 of 28 Linguistic Variable The linguistic variable I will be focussing on in this study is -t/-d deletion (consonant cluster simplification) in two varieties of English, describing the major linguistic constraints that govern variation. TD deletion in English refers to the deletion of the alveolar stop /t/ and /d/ in final consonant clusters, although in this study I will also briefly look at deletion in word medial position. This project will look at two important factors regarding this deletion. Linguistic Factor - the phonological environment the t/d occurs in and the morpheme it occurs in Social Factor - The age and sex of a speaker, their social class and speaking style. I will look at three linguistic factors that influence (either by promoting or disfavouring) deletion . Two of these factors consider the phonological environment that the -t/-d occurs and the other considers the grammatical category, in other words the type of morpheme that the -t/-d cluster occurs in. The Preceding Segment This refers to the class of sounds that occur before the -t/-d, basically meaning the first part of the consonant cluster (and therefore it is limited to consonants.) These can be divided into five groups 1) Sibilants / j, s, ʃ, ʧ, z,ʤ / 2) Stops /b, p, g, k/ 3) Fricatives /f, v, θ, ð/ 4) Nasals /n, m, ŋ/ 5) Liquids /l/ E.g. I passed the exam, I was parched for a drink. E.g. I stopped the car, E.g. I left home E.g. Send it in the post. E.g I told him. The Following Segment This refers to the class of sounds that follow the t/-d cluster such as a morpheme or pause etc. We can also divide these into five groups: 1) Consonants 2) Glides 3) Rhotics 4) Vowel 5) Pause This refers to all the categories above /y/, /w/, /h/ /r/ (All English vowels) Silence, break in speech. The Grammatical Category The grammatical category refers to the type of morpheme that the t/d occurs in. I have out these into four categories: 1) Mono-morphemes 2) Irregular devoicing verbs 3) Regular past verbs 4) Semi weak verbs Where the t/d is part of the stem. E.g. Just E.g. send/sent E.g. She passed the salt. E.g. I left the cooker on. With regard to grammatical category, I have excluded negative suffix items such as can’t and won’t from my data as I felt that these forms were fairly predictable in the way they work and that they are the most deleted forms. page 7 of 28 Exclusions from my Data In my project I have excluded some specific examples which I feel cannot provide an accurate account of the deletion of -t/-d in these two dialects. I have excluded the following cases: Cases where a final -t/-d is followed by a -t/-d (respectively) at the beginning of the next morpheme. This final -t/-d is deleted, almost in anticipation of the following -t/-d. E.g. It's just too much. Cases where a nasal appears before a -t/-d cluster and where a vowel follows (the rule of nasalflapping) E.g. Want a light becomes wanna light. Cases where preceding /r/ is a vital part of the consonant cluster. E.g. This is part of it. Cases of the function word and as this is very predictable in how its work and therefore there was no need to include it in my data. Major Constraints According to Labov (1989) the major constraints on –t/-d deletion goes as follows. I will use these formulas in more detail when I discuss my Results. Theoretical sonority hierarchy: Stops > Fricatives > Sibilants > Nasals > /l/ LOW SONORITY………………………………………………… Preceding segment: Deletion Rule: HIGH SONORITY Sibilant > Stops > Nasals > Fricatives > /l/ Most favoured………………………Least favoured Following segment: Consonant > Liquid, Glide > Vowel > Pause Deletion Rule: Most favoured………………………Least favoured Grammatical status: Mono-morpheme > Semi weak verbs > Irregular devoiced verbs > Regular past tense Deletion Rule: Most favoured………………………Least favoured page 8 of 28 Methods In this project I have looked at two different areas, Bridlington (Yorkshire) and Colchester (Essex). I chose to look at two areas, so I could compare my findings with another another of Britain and in particular look at whether there were any variation between the North and the South of England. I chose the Bridlington sample as I have some previous knowledge of the Yorkshire accent, having lived there, and the Colchester sample because this is where I currently live. I have used a sample of seven speakers in total, five speakers from Bridlington and two speakers from Colchester. The Bridlington sample consisted of one old male, two old females, and two young females. The Colchester sample consisted of an old male, and a young male. All the data collections were taken from informal interviews and I felt this was important in obtaining an accurate account of the features of these speakers. I chose to look at speakers of both sexes and a range of ages so I could look at whether there was a language change in progress and to see if there were any distinctive differences between the two areas. The Bridlington sample only provided us with the gender and whether the speaker was old or young. The Colchester sample provided us with many more sociolinguistic factors such as where they’ve lived and their social class. I felt it was important to include a variety of speakers in order to analyse the data and come to accurate conclusion. The table below shows the sample I used and some relevant information that was provided with the sample. Tape Reference Brid04 Brid04 Brid05 Brid11 Brid12 457.1.98 457.13.98 Name of subject Unknown Unknown Unknown Unknown Unknown Vic Bagnall Darren Parker Age Old Old Old Young Young d.o.b 13/10/44 d.o.b. 1969 Gender Male Female Female Female Female Male Male Sample Bridlington Bridlington Bridlington Bridlington Bridlington Colchester Colchester To calculate my results, I extracted between 30 and 50 tokens per person, depending on the quality of the tape and the amount of tokens I could actually find by a particular speaker. I looked at three linguistic factors (mentioned in the Linguistic Variable), two of which consider the phonological environment that the -t/-d occurs in and the other considers the grammatical category, in other words the type of morpheme that the -t/-d cluster occurs in. The following are all examples of the kind of phonological environments I looked at in my project to see if deletion occurred: Where a consonant cluster ends and a morpheme boundary or pause follows: t, d ___ E,g I left. Where a consonant cluster ends and a vowel follows: t, d ___V E.g I left it on. Where a consonant cluster ends and a consonant follows t, d ___C E.g I left the cooker on. page 9 of 28 How past tense marking effects consonant cluster deletion E.g. I passed the exam. I displayed my results firstly in the form of tables, showing each speakers, preceding segment and following segment and then the grammatical category the –t/-d was found in. The tables display the number of tokens found in each speakers recording and then I have calculated the percentages so we can see a clearer picture. (My full results can be seen in both the Results and Appendix sections.) I then displayed all my results in the form of graphs which can be seen in the Results section. In this section, I review my findings and explain what my recorded data shows, describing in the Conclusion what their relevance is to previous studies carried out on this feature. page 10 of 28 Results: Colchester Old Male 457.1.98 (Vic Bagnall) Percentage of deleted -t/-d tokens by Preceding Segment Preceding Segment Sibilant Stop Fricative Nasal Liquid No. of Tokens 29 0 0 19 2 Percentage 58% 0% 0% 19% 4% Percentage of deleted -t/-d tokens by Following Segment Following Segment Consonant Glide Rhotic Vowel Pause No.of Tokens 46 1 2 1 0 Percentage 92% 2% 4% 2% 0% Percentage of deleted -t/-d tokens by Grammatical Category Grammatical Category Mono-morphemes Semi weak Irregular devoicing Regular past Medial 3 Consonant cluster No.of Tokens 32 3 0 0 1 14 Percentage 64% 6% 0% 0% 2% 28% The first graph for this speaker shows that the first constraint (preceding segment) is clearly an important factor that determines whether the following -t/-d is deleted, favouring that sibilants influence deletion the most, it represents 58% of the deleted tokens. The second constraint (following segment) showed that deletion was most favoured when followed by a consonant which was expected. The final constraint shows that mono-morphemes clearly favour deletion as I found no tokens that show deletion in Irregular devoiced verbs or Regular past tense. There were a small number of semi weak verbs that underwent deletion but too small to make any concrete conclusions. The second most favoured place for deletion was where there were a string of three consonants in the final cluster. We can therefore translate our data into a hierarchy for each constraint. Preceding: Sibilants > Nasals > Liquids > Stops > Fricatives Following: Consonants> Rhotics > glides > Vowels > Pause Category: Mono > Semi weak > Irreg devoicing > Reg past page 11 of 28 Results: Colchester Young Male 457.13.98 (Darren Parker) Percentage of deleted -t/-d tokens by Preceding Segment Preceding Segment Sibilant Stop Fricative Nasal Liquid No. of Tokens 17 1 0 8 0 Percentage 66% 4% 0% 30% 0% Percentage of deleted -t/-d tokens by Following Segment Following Segment Consonant Glide Rhotic Vowel Pause No.of Tokens 20 2 4 0 0 Percentage 77% 8% 15% 0% 0% Percentage of deleted -t/-d tokens by Grammatical Category Grammatical Category Mono-morphemes Semi weak Irregular devoicing Regular past Medial 3 Consonant cluster No.of Tokens 21 0 0 0 1 4 Percentage 81% 0% 0% 0% 4% 15% The first constraint, as with the older Colchester male, shows that deletion is favoured by sibilants the most and then nasals follow, 66% and 30% respectively. This is where the similarity ends as the young male appears to favour deletion more when a stop precedes than when a liquid or fricative. There is only one token to show that stops are favoured more so it could be the case that with a larger sample I would not find this. The second constraint shows the same results as that of the older male, in that consonants favour deletion the most. The grammatical category constraint also shows the same findings. The majority of tokens are deleted in mono-morphemes and those that were not are either word medial or part of three consonants in the cluster. I was unable to find any tokens that showed deletion in semi weak verbs, Irregular devoiced verbs or regular past endings. This is the hierarchy for the Darren Parker. Preceding: Sibilants > Nasals > Stops > Fricatives, Liquids Following: Consonants > Rhotics > Glides > Vowel, Pause Category: Mono > Semi weak, Irreg devoicing, Reg past page 12 of 28 Results: Bridlington Old Male 04 Percentage of deleted -t/-d tokens by Preceding Segment Preceding Segment Sibilant Stop Fricative Nasal Liquid No. of Tokens 14 2 2 17 1 Percentage 39% 6% 6% 46% 3% Percentage of deleted -t/-d tokens by Following Segment Following Segment Consonant Glide Rhotic Vowel Pause No.of Tokens 34 0 1 0 1 Percentage 94% 0% 3% 0% 3% Percentage of deleted -t/-d token by Grammatical Category Grammatical Category Mono-morphemes Semi weak Irregular devoicing Regular past Medial 3 Consonant cluster No.of Tokens 23 2 0 4 0 7 Percentage 64% 5.5% 0% 11% 0% 19.5% The Bridlington results show some interesting differences from those found in the Colchester corpus data. The first constraint shows that deletion is favoured more when the preceding segment is a nasal. There is a 13% difference between that and sibilants, which come as the second most favoured but this is a little misleading as there are only three tokens separating these results. The second constraint shows that following consonants are favoured for deletion 94% of the time by this speaker. This therefore leaves only a small percentage for the other factors and we can conclude that deletion is most favoured by consonants. The final constraint shows that mono-morphemes are favoured most for deletion followed by a very surprising regular past endings (11%) coming as second most favoured for deletion. This will be discussed further in the Conclusion. Preceding: Nasals > Sibilants > Stops > Fricatives > Liquids Following: Consonants > Rhotics > Pause > Glides, Vowels Category: Mono > Reg past > Semi weak > Irreg devoicing page 13 of 28 Results: Bridlington Old Female 04 Percentage of deleted -t/-d tokens by Preceding Segment Preceding Segment Sibilant Stop Fricative Nasal Liquid No. of Tokens 24 0 3 2 2 Percentage 78% 0% 10% 6% 6% Percentage of deleted -t/-d tokens by Following Segment Following Segment Consonant Glide Rhotic Vowel Pause No.of Tokens 25 5 0 1 0 Percentage 81% 16% 0% 3% 0% Percentage of deleted -t/-d tokens by Grammatical Category Grammatical Category Mono-morphemes Semi weak Irregular devoicing Regular past Medial 3 Consonant cluster No.of Tokens 26 1 0 3 1 0 Percentage 84% 3% 0% 10% 3% 0% The Bridlington Female 04 results show that she favours deletion the most when the preceding environment is a sibilant, at a clear advantage of 78% of the deleted tokens. This is then followed by Fricatives. The second constraint clearly shows that following consonants favour deletion the most, as found with all the other speakers so far and represents 81% of the deleted tokens. The final constraint shows that mono-morphemes represented 81% of the deleted tokens but few tokens were found for regular past, semi weak and irregular devoiced verbs. As with the Bridlington Old Male 04, 10% of the deleted tokens were from regular past tense endings and again this will discussed further in the Conclusion. Preceding: Sibilants > Fricatives Nasals, Liquids > Stops Following: Consonants > Glides > Vowels > Rhotics > Pause Category: Mono > Reg past > Semi weak, Irreg devoicing page 14 of 28 Results: Bridlington Old Female 05 Percentage of deleted -t/-d tokens by Preceding Segment Preceding Segment Sibilant Stop Fricative Nasal Liquid No. of Tokens 28 0 0 7 1 Percentage 78% 0% 0% 19.5% 2.5% Percentage of deleted -t/-d tokens by Following Segment Following Segment Consonant Glide Rhotic Vowel Pause No.of Tokens 24 7 2 1 2 Percentage 66% 20% 5.5% 3% 5.5% Percentage of deleted -t/-d tokens by Grammatical Category Grammatical Category Mono-morphemes Semi weak Irregular devoicing Regular past Medial 3 Consonant cluster No.of Tokens 31 0 1 0 1 3 Percentage 86% 0% 3% 0% 3% 8% The Bridlington Female 05 results show some different findings from the other Bridlington Old female. She favours deletion with a preceding sibilant the most, both representing 78% of the deleted tokens. Although the hierarchy below looks quite different to that of the other old female, percentages are not very different. The only main difference with this constraint is the positioning of fricatives and liquids which I shall discuss further in my Conclusion. The second constraint shows that consonants are favoured but only 66% of the time. Glides represent another fairly substantial amount with 20% of the deleted tokens. The final constraint clearly shows, with 86% that this is the most favoured. I did, however, find no tokens to represent deleting of regular past tense endings. The hierarchy therefore goes as follows: Preceding: Sibilants > Nasals > Liquids > Fricatives, Stops. Following: Consonants > Glides > Rhotics > Pause > Vowel Category: Mono > Irreg devoicing > Semi weak, Reg past page 15 of 28 Results: Bridlington Young Female 11 Percentage of deleted -t/-d tokens by Preceding Segment Preceding Segment Sibilant Stop Fricative Nasal Liquid No. of Tokens 21 0 1 8 0 Percentage 70% 0% 3% 27% 0% Percentage of deleted -t/-d tokens by Following Segment Following Segment Consonant Glide Rhotic Vowel Pause No.of Tokens 23 7 0 0 0 Percentage 77% 23% 0% 0% 0% Percentage of deleted -t/-d tokens by Grammatical Category Grammatical Category Mono-morphemes Semi weak Irregular devoicing Regular past Medial 3 Consonant cluster No.of Tokens 22 1 0 0 3 4 Percentage 74% 3% 0% 0% 13% 10% This speakers results show that she favours deletion 70% of the time when the preceding environment is a sibilant. The next factor was nasals which consisted of 27% of the total tokens. The second constraint showed that 77% of the deleted tokens were consonants, followed by glides with 23%. The data shows that I found no tokens of any other following segment. The final constraint shows that 77% of the deleted tokens were mono-morphemes and 3% of them were semi weak verbs. The other 23% consisted of word medial tokens and consonant clusters containing three consonants. I have excluded the latter two for reasons I will discuss in the Conclusion. The hierarchy therefore goes as follows: Preceding: Following: Category: Sibilants > Nasals >Fricatives > Stops, Liquids Consonants > Glides > Rhotics, Vowels, Pause Mono > Semi weak > Irreg devoicing, Reg past page 16 of 28 Results: Bridlington Young Female 12 Percentage of deleted -t/-d tokens by Preceding Segment Preceding Segment Sibilant Stop Fricative Nasal Liquid No. of Tokens 11 0 0 9 0 Percentage 55% 0% 0% 45% 0% Percentage of deleted -t/-d tokens by Following Segment Following Segment Consonant Glide Rhotic Vowel Pause No.of Tokens 17 3 0 0 0 Percentage 85% 15% 0% 0% 0% Percentage of deleted -t/-d tokens by Grammatical Category Grammatical Category Mono-morphemes Semi weak Irregular devoicing Regular past Medial 3 Consonant cluster No.of Tokens 14 0 0 0 0 6 Percentage 70% 0% 0% 0% 0% 30% For this speaker, the results show that 55% of the deleted tokens favour deletion when the preceding segment is a sibilant. This is lower than that found in the other young Bridlington speaker. She then favours the deletion when preceded by a nasal 45%. I found for the following environment, exactly the same hierarchy as that of the other young Bridlington female, with 85% of the tokens favouring deletion before a consonant. For the third constraints I found no tokens other than mono-morphemes, which represented 70% of my tokens. I found no other tokens that represented semi weak verb, irregular devoiced verbs or regular past tense endings. The other 30% of the deleted tokens were from consonant clusters that contained three consonants. The hierarchy therefore goes as follows: Preceding: Following: Category: Sibilants > Nasals > Stops, Fricative, Liquids. Consonants > Glides > Rhotics, Vowels, Pause Mono > Semi weak, Irreg devoicing, Reg past page 17 of 28 Results for Bridlington Old Male 04 Percentage of deleted -t/-d tokens by preceding segment 50 45 40 35 30 % 25 20 15 10 5 0 Sibilant Stop Fricative Nasal Liquid Percentage of deleted -t/-d tokens by following segment 100 90 80 70 60 % 50 40 30 20 10 0 Consonant Glide Rhotic Vowel Pause Percentage of deleted -t/-d tokens by grammatical category 70 60 50 40 % 30 20 10 0 mono Semi Irreg Reg past Medial 3 cons page 18 of 28 Results for Bridlington Old Female 04 Percentage of deleted -t/-d tokens by preceding segment 80 70 60 50 % 40 30 20 10 0 Sibilant Stop Fricative Nasal Liquid Percentage of deleted -t/-d tokens by following segment 90 80 70 60 50 % 40 30 20 10 0 Consonant Glide Rhotic Vowel Pause Percentage of deleted -t/-d tokens by grammatical category 90 80 70 60 50 % 40 30 20 10 0 Mono Semi Irreg Reg past Medial page 19 of 28 Results for Bridlington Old Female 05 Perecentage of deleted -t/-d tokens by preceding segment 80 70 60 50 % 40 30 20 10 0 Sibiant Stop Fricative Nasal Liquid Percentage of deleted -t/-d tokens by following segment 70 60 50 40 % 30 20 10 0 Consonant Glide Rhotic Vowel Pause Percentage of deleted -t/-d tokens by grammatical category 90 80 70 60 50 % 40 30 20 10 0 Mono Semi Irreg Reg past Medial 3 cons page 20 of 28 Results for Bridlington Young Female 11 Percentage of deleted -t/-d tokens by preceding segment 70 60 50 40 % 30 20 10 0 Sibilant Stop Fricative Nasal Liquid Percentage of deleted -t/-d tokens by following segment 80 70 60 50 % 40 30 20 10 0 Consonant Glide Rhotic Vowel Pause Percentage of deleted -t/-d tokens by grammatical category 80 70 60 50 % 40 30 20 10 0 Mono Semi Irreg Reg past Medial 3 cons page 21 of 28 Results for Colchester Old Male 457.1.98 (Vic Bagnall) Percentage of deleted -t/-d tokens by preceding segment 60 50 40 % 30 20 10 0 Sibilant Stop Fricative Nasal Liquid Percentage of deleted -t/-d tokens by following segment 100 90 80 70 60 % 50 40 30 20 10 0 Consonant Glide Rhotic Vowel Pause Percentage of deleted -t/-d tokens by grammatical category 70 60 50 40 % 30 20 10 0 Mono Semi Irreg Reg past Medial 3 cons page 22 of 28 Results for Colchester Young Male 457.13.98 (Darren Parker) Percentage of deleted -t/-d tokens by preceding segment 70 60 50 40 % 30 20 10 0 Sibilant Stop Fricative Nasal Liquid Percentage of deleted -t/-d tokens by following segment 80 70 60 50 % 40 30 20 10 0 Consonant Glide Rhotic Vowel Pause Percentage of deleted -t/-d tokens by grammatical category 90 80 70 60 50 % 40 30 20 10 0 Mono Semi Irreg Reg past Medial 3 cons page 23 of 28 Results for Bridlington Young Female 12 Percentage of deleted -t/-d tokens by preceding segment 60 50 40 % 30 20 10 0 Sibilant Stop Fricative Nasal Liquid Percentage of deleted -t/-d tokens by following segment 90 80 70 60 50 % 40 30 20 10 0 Consonant Glide Rhotic Vowel Pause Percentage of deleted -t/-d tokens by grammatical category 70 60 50 40 % 30 20 10 0 Mono Semi Irreg Reg past Medial 3 cons page 24 of 28 Summary of Results In this section of the paper, we saw the results of my recorded data through a series of tables and graphs. The tables show the number of deleted tokens I found from each speaker and then they have been calculated as percentages to give a clearer picture. In this section, we will summarise the observations of the data, discussing each speaker individually and then the group as a whole. Due to the very small sample of Colchester data, the results cannot completely account for the way t/-d operates in Colchester. These results can draw some reasonable conclusions but these may be speculative and with more speakers, we could study this area further. My results show that both Colchester speakers favour sibilants and then nasals in the preceding segments for deletion. The hierarchy then shows mixed results but only three tokens separate these findings and so no concrete conclusions can be made with such few tokens between them. Both show a high percentage of deleted -t/-d tokens when the following segment is a consonants, however the younger speaker shows that he is more likely to delete when there is a following rhotic or glide than the older male. For the third constraint, both showed the highest percentage of deletion with mono-morphemes. Four out of the five Bridlington speakers favoured sibilants as the most likely preceding environment in which to delete a /t/ or /d/. The Bridlington Male was the only exception but the result was inclusive as only three tokens separated nasals and sibilants. The two old females both had 78% deleted tokens as sibilants for the first constraint. The second constraint showed some difference between the two older females, both favouring consonants but one having 81% deleted tokens and the other 66%. The third constraint showed that both speakers favoured mono-morphemes for deletion. The two young females showed a difference in percentage of deleted sibilants, with one showing 70% of deleted sibilants and the other 55%. For the second constraint, both had a high number of deleted consonants, recording 77% and 85% of the tokens that were collected. For the final constraint both speakers favoured mono-morphemes the most for deletion. This paper showed that six out of the seven speakers favoured a preceding sibilant the most for deletion before a -t/-d. The only exception to this data was the Bridlington Old Male 04, who favoured nasals the most with 46% and sibilants following with 39%. The percentages indicate that there is some difference between the two but in fact, there are only three tokens between nasals and fricatives and so with a larger sample we may see a different pattern which follows that of the other speakers. For the second constraint (the following environment) the data showed that all the speakers favoured consonants the most for deletion, although the range expanded from 66% (Old Female 05) to 94% (Bridlington Old Male 04.) For the third constraint all the speakers again favoured mono-morphemes as the likeliest to delete. page 25 of 28 Conclusion I can conclude several things from the sample of data I studied in this project. My hypothesis was fairly correct but there were some subtle differences between the two areas and the individuals. I found that the phonological environment is very important in influencing the deletion of /t/ and /d/, the biggest influence being that of the following segment. The linguistic context, without a doubt, effects whether or not deletion occurs. Previous research and my results both clearly show that deletion is far less likely before vowels than before consonants. From my results I can propose a hierarchy of the likeliness of deletion. Most likely Less likely 1. Preceding consonant and following consonant E.g. Best friend, cold weather 2. Past tense suffix and following consonant E.g. He passed the salt, she seemed funny 3. Preceding consonant and following vowel E.g. Most of the time, ground attack 4. Past tense suffix and following vowel. E.g. I passed out, she seemed odd. Consonant Clusters Some of my examples show lenition, a process which is where a consonant is weakened in certain positions. T/D deletion of this sort occurs in examples such as friends, hands and bands show lenition in certain positions. Notice that most of these examples contain a /d/ rather than a /t/ and that it is deleted due to it being the least salient sound in the consonant cluster. Where this occurs with /t/, however, it is not deleted but infact pronounced as a glottal stop. This also occurs in some word medial positions such as in ammendment and commitment where the /d/ is deleted and the /t/ pronounced as a glottal stop respectively. Past Tense Marking As far as past tense marking occurs, my two Colchester speakers have no tokens that show that they delete –t/-d when marking the past tense. I found that two out of five of my Bridlington speakers actually do delete sometimes when marking the past tense. The numbers of these were quite low and may be due to Yorkshire Assimilation, a process whereby a final /d/ becomes a /t/. For example bedtime would become bettime. Previous Studies Reviewing what previous studies have found, my results do not show much difference in results from what they observed although most of these studies were sociolinguistic as opposed to phonological. I found similar patterns in the factors that favoured deletion Labov’s Hierarchy 1968 (shown in the Linguistic Variable) seems to support my findings on the most favoured environments but not on the following. The only real major difference between my findings and Labov's hierarchy is the positioning of stops, as nasals appear to be more favoured across the speakers. Discussion There are a few important points which should be mentioned in relation to my data and the results. The first is that these results show the number of deleted tokens within each speakers recording but do not calculate the actual total number of tokens which contain consonant clusters. This therefore shows that this data does not fully represent the patterns that are occurring in either area as the percentage of absence tokens was not calculated. Although these results confirm the factors that are most favoured page 26 of 28 for deletion, by calculating the percentage of absence tokens we would get a much clearer and accurate picture. The data shows that fricatives are least favoured for deletion when they precede a /t/ or /d/. This is obvious to conclude as /θ/ and /ð/ cannot occur before a /t/ or /d/ consonant cluster and /v/ does not occur either. This therefore only leaves /f/ and these tokens are small. Also that mono-morphemes were most favoured for deletion but if we look at the other grammatical categories we can see that this should clearly be the case. Semi weak verbs are very rare , Irregular verbs are not very common and Regular past tense endings should be kept and not deleted as they are the only indication of the past tense of the verb. It would be interesting to expand this study with more speakers and certainly more tokens in order to get a clearer picture. Some of the tapes were poor quality and I would like to conduct further interviews myself as I feel the interviewer has the best angle at transcribing and concluding the data. I would also like to expand my study to other areas of Britain to see if any further variation could be found page 27 of 28 References Books Cruttenden, A. (2001) Gimson’s Pronunciation of English Arnold, London Foulkes and Docherty (1999) Urban Voices. Arnold; London. Patrick, P.L (1999) Urban Jamaican Creole: Variation in the mesolect. John Benjamins; Amsterdam & Philadelphia Radford et al ( 1999) Linguistics: An Introduction. CUP; Cambridge. Wells, J.C (1982) Accents of English. Vol 1 – Introduction. CUP; Cambridge. Wells, J.C (1982) Accents of English.Vol 2 – The British Isles. CUP; Cambridge.Journal Articles Journal Articles Bayley, R (1994: 303-326) Consonant cluster simplification in Tejano English. Journal of Language Variation and Change 6. Guy, G (1991: 1-22) Explanation in variable phonology: An exponential model of morphological class. Journal of Language Variation and Change 3. Guy, G (1991: 223-239) Contextual conditioning in variable lexical phonology. Journal of Language variation and Change 3 Labov, W (1989: 85-97) The child as linguistic historian in Journal of Language Variation and Change 1. Patrick, P.L (1991: 171-189) Creole at the intersection of variable processes: -t, d deletion and past-marking in Jamaican mesolect in Journal of Language Variation and Change 3. Santa Ana A, O (1996: 63-89) Sonority and syllable structure in Chicano English. Journal of Language Variation and Change 8 page 28 of 28 Websites www.ling.upenn.edu – Daniel Schreier article on Maori English www.ling.osu.edu - Word-medial /t, d/ deletion in spontaneous speech Acknowledgements Roger Jennings LG232 Lecture Notes – Introduction to Sociolinguistics Wyn Johnson and Dave Britain