Levelling and diffusion in the Cumbrian city dialect of Carlisle 51 Edited by Rainer Schulze Advisory Board: Anita Fetzer (Augsburg), Thomas Herbst (Erlangen), Andreas Jucker (Zürich), Manfred Krug (Bamberg), Christian Mair (Freiburg i.Br.), Ute Römer (Atlanta, GA, USA), Andrea Sand (Trier), Hans-Jörg Schmid (München), Josef Schmied (Chemnitz) and Edgar W. Schneider (Regensburg) Sandra Jansen Levelling and diffusion in the Cumbrian city dialect of Carlisle Bibliografische Information der Deutschen Nationalbibliothek Die Deutsche Nationalbibliothek verzeichnet diese Publikation in der Deutschen Nationalbibliografie; detaillierte bibliografische Daten sind im Internet über http: / / dnb.dnb.de abrufbar. © 2018 · Narr Francke Attempto Verlag GmbH + Co. KG Dischingerweg 5 · D-72070 Tübingen Das Werk einschließlich aller seiner Teile ist urheberrechtlich geschützt. Jede Verwertung außerhalb der engen Grenzen des Urheberrechtsgesetzes ist ohne Zustimmung des Verlages unzulässig und strafbar. Das gilt insbesondere für Vervielfältigungen, Übersetzungen, Mikroverfilmungen und die Einspeicherung und Verarbeitung in elektronischen Systemen. Printed in Germany Internet: www.narr.de E-Mail: info@narr.de ISSN 0939-9399 ISBN 978-3-8233-9036-7 1 7 2 9 2.1 9 2.1.1 10 2.1.2 14 2.1.3 15 2.2 20 2.3 20 3 23 3.1 23 3.2 24 3.3 26 3.4 29 4 31 4.1 31 4.2 32 4.3 33 4.4 35 4.5 36 4.6 37 4.6.1 38 4.6.2 39 5 41 5.1 41 5.1.1 42 5.1.2 45 5.1.3 47 Contents Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Theoretical framework . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Dialect contact . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Accommodation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Levelling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Geographical diffusion . . . . . . . . . . . . . . . . . . . . . . . . . . Regional identity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Dialects in contact: The Carlisle setting . . . . . . . . . . . . . . . . . . Carlisle and its geographic, historic and sociolinguistic background . . . . Geography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Demographics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . History . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Carlisle in the northern English context . . . . . . . . . . . . . . . . . . Methodology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Age as external factor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Sex as external factor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Social class as external factor . . . . . . . . . . . . . . . . . . . . . . . . . . . Sampling methods and contacting speakers . . . . . . . . . . . . . . . Final sample . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Interviews . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Interview procedure and oberserver’s paradox . . . . . . . Style . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Sociolinguistic variables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . G O O S E , G OAT and F O OT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . G OO S E , G OAT and F OOT realisation in England . . . . . . . . Data analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Normalisation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.1.4 48 5.1.5 49 5.1.6 50 5.1.7 59 5.1.8 61 5.2 62 5.2.1 62 5.2.2 64 5.2.3 67 5.2.4 67 5.2.5 68 5.2.6 79 5.2.7 81 5.3 82 5.3.1 83 5.3.2 85 5.3.3 85 5.3.4 86 5.3.5 102 5.3.6 104 5.4 104 5.4.1 105 5.4.2 108 5.4.3 108 5.4.4 109 5.4.5 109 5.4.6 119 5.4.7 121 6 122 7 129 131 144 145 147 Classification of fronting . . . . . . . . . . . . . . . . . . . . . . . . . Statistical modelling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (T) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Historical and geographical background . . . . . . . . . . . . (T) in Britain . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Data analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Statistical modelling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (R) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Historical and geographical background . . . . . . . . . . . . Data analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Statistical modelling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (TH) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Historical and geographical background . . . . . . . . . . . . Phonetic details . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Data analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Statistical modelling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . List of Figures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . List of Tables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Contents 6 1 Introduction For its novelty to the newcomer, its variety and its preserving so much that is historical, Carlisle’s must surely rank as one of the most interesting city dialects in the British Isles (Wright 1978: 15) This is a sociophonetic study of speakers from Carlisle, Cumbria in the far north west of England, an area which has had little attention from a dialectological and variationist sociolinguistic point of view. Language change in the com‐ munity is studied by investigating variation in the speech of adults of different ages. The results are set in the context of geolinguistic processes, i.e. geograph‐ ical diffusion and dialect levelling. These are mechanisms in dialect contact sit‐ uations which lead to language change. The thesis examines six phonological features of Carlisle English in a varia‐ tionist sociolinguistic framework. Auditory and acoustic methods as well as statistical modelling are employed to investigate processes of language change. The following research questions are addressed: • Which geolinguistic processes are observable? • How does linguistic variation pattern across the social categories age, sex and social class in Carlisle English? • Is variation leading to language change? If so, are these changes led by internal or external motivation? • Can geolinguistic processes explain variation and change in the linguistic system of Carlisle English? The book is structured as follows: Chapter 2 presents the theoretical framework of this study. I concentrate on a dialect contact framework since dialect contact (often due to mobility) is re‐ sponsible for various changes that have occurred and are occurring in varieties across the north of England. Of considerable interest is Trudgill’s (1986) seminal work Dialects in Contact. Processes such as geographical diffusion and dialect levelling are identified as key processes in dialect contact scenarios. Moreover, external an internal factors of language change are discussed. Chapter 3 introduces Carlisle as research site comprising essential geo‐ graphic, demographic and historic background information. Chapter 4 provides a detailed account of the data collection and highlights relevant issues in connection with the fieldwork. In chapter 5 the variation of the following phonetic variables is discussed: The first of these variables is the degree of fronting of the G O O S E vowel. In addition, results of possible fronting of G OAT and F O OT are discussed in relation to the G O O S E vowel. For studying variation and change in Carlisle English vowels, acoustic phonetics as a mean of measuring variation has been chosen. The use of (T) in intervocalic and word-final position such as butter or hat but not in word-initial position such as tea is analysed. In contrast to the vowel variables, the consonant variables are studied by auditory means. The use of (R) in words such as red, merry and free is considered. However, / r/ in final or pre-consonantal position (rhoticity) such as pair or park are not taken into account here. The use of voiced and voiceless variants of (TH) in words such as other, breathe, three, something and north is analysed. The sociolinguistic variables have been selected because of their ubiquity in northern English varieties due to diffusional mechanisms (e.g. Burbano-Eliz‐ ondo 2006; Llamas 2007; Atkinson 2011; Flynn 2012). One goal is to observe whether these features are already part of Carlisle English despite Carlisle’s geographical distance from London. Another goal will be to see if a change in progress is observable. The diffusional mechanisms within the community but also in relation to other urban areas in close proximity to Carlisle will be taken into account. In chapter 6 the results obtained in chapter 5 are discussed and common themes are identified. They are related to theories and findings in the literature on language variation and change. The research objectives proposed in this chapter are revisited in the light of the results. In addition, the results are linked to social practice in Carlisle English. The analysis of social practice of people can yield important insights into the linguistic behaviour. The conclusion in chapter 7 completes the book. 1 Introduction 8 1 In contrast to language contact scenarios that can lead to different outcomes such as bilingualism, diglossia, pidgins and creoles. 2 Theoretical framework This thesis draws on sociophonetic methods. The study provided here describes variation and change processes in Carlisle English. Dialect contact is an impor‐ tant factor in dialect change and this is the starting point for this study. I inves‐ tigate changes in Carlisle English which are due to dialect contact scenarios, in particular the geolinguistic processes diffusion and levelling. 2.1 Dialect contact This part of the chapter deals with the dialect contact framework. Dialect contact due to mobility lies at the heart of the geolinguistic processes diffusion and levelling which have been discussed widely in the sociolinguistic literature in recent years (e.g. Trudgill 1986; Britain 2002; Kerswill 2003; Steele 2008; At‐ kinson 2011; Flynn 2012). Mobility increased dramatically over the course of the 20 th century and is a major part of today’s society. Increased mobility leads to an increase in contact between individuals from different geographical and social areas and thus to the increase of dialect contact scenarios which Trudgill (1986: 1) defines as “contact between varieties of language that are mutually intelligible.” 1 Possible contact scenarios are when people go on holiday or when they travel to the next bigger city for shopping or for cultural events. These are rather short-term contacts and are often discarded in dialect contact research while commuting and moving can be interpreted as long-term contact situations which can lead to and further language change. But not only mobility can influence language change. The area that people orient themselves towards or away is also an important factor (cf. Montgomery 2012; Leach, Watson and Gnevsheva 2016). Britain (2010b) em‐ phasises social practice, attitudes and orientation as key factors for dialect change scenarios. In the following part I present various processes that are related to dialect contact scenarios. 2.1.1 Accommodation Accommodation has to be seen as a supposition in dialect contact scenarios. It refers to the linguistic changes on the individual level in short-term contact situations. In general, speakers tend to modify their speech towards or away from their interlocutor. This hypothesis is put forward by Giles and Powesland (1975 quoted in Trudgill 1986) in the communication accommodation theory. Their explanation for such behaviour is similar to positive politeness strategies found on the pragmatic level. The essence of the theory of accommodation lies in the social psychological research on similarity-attraction. This work suggests that an individual can induce another to evaluate him more favourably by reducing dissimilarities between them. The process of speech accommodation operates on this principle as such may be a reflection of an individual’s desire for social approval (Giles and Powesland 1975: 157). Studies have shown short-term accommodation of speakers towards the inter‐ locutor. Coupland (1984) demonstrates in his study that a travel agent accom‐ modates her speech towards her customers. Bell (1984) on the other hand anal‐ yses the accommodation of radio presenters to their invisible audience (an approach which is also called audience design). However, short-term accommo‐ dation and long-term accommodation have to be distinguished. Auer (2007: 109) restricts the impact of short-term accommodation when he states that “it cannot be taken for granted that the social psychological model which explains inter‐ personal accommodation can be expanded to explain long-term dialect accom‐ modation as well.” Thus, he argues that while short-term accommodation is a non-permanent change, long-term accommodation might lead to changes that last and modify the original variety considerably. Trudgill (1986: 24) claims that long-term accommodation follows a fixed route, i.e. speakers with a similar linguistic background acquire features of an‐ other linguistic background in a similar manner. He bases this assumption on data of English citizens acquiring American English features as a result of long-term accommodation. Trudgill ascribes an important role to salience in the accommodation process which he defines as “due to factors such as [...] stig‐ matization, linguistic change, phonetic distance and phonological contrast” (Trudgill 1986: 11). A predecessor to the concept of salience is Schirmunski’s (1928/ 29 and 1930 quoted in Schwarz, Spiekermann and Streck 2011) model of primary and secondary dialect features. Schirmunski argues that primary dialect features are more ostentatious than secondary dialect features which lead to the loss of the former ones in dialect contact situations (see Lenz 2010 for a historical overview on the notion of salience). 2 Theoretical framework 10 However, the concept of salience is highly debatable. Meyerhoff (2011) criti‐ cises the term salience as “a maddeningly under-defined term when used in sociolinguistics” and different approaches to define salience exist: Hickey (2000) provides a list of possible triggers for salience. He does however emphasise that none of these factors can be given more importance than others and that they can co-occur. His list of factors includes acoustic prominence, homophonic merger, system conformity, deletion and insertion, grammatical restructuring, openness of word class, the loss of vernacular features and retention of condi‐ tional realisations. Hickey (2000: 13) suggests that “the features involved are always strongly local and contrast with their lack in varieties which are less regionally bound, more ‘standard’ in the sense that they are not primarily in‐ dicative of a specific geographical area or of a social group.” Kerswill and Williams (2002) introduce a model which takes salience into account as a possible factor in language change. Social factors as well as lan‐ guage-internal factors are important in their approach to explain salience, e.g. stark phonetic contrast. However, internal factors are only a prerequisite for features to become salient. Auer, Barden and Grosskopf (1998) observe changes in the speech of individ‐ uals in a long-term dialect contact situation. They choose various factors men‐ tioned by Schirmunski and Trudgill and divide them into subjective and objec‐ tive aspects: articulatory distance, areal distribution, phonemicity, continuous vs. dichotomous structure, lexicalisation as objective criteria and code-switching, representation in writing and stereotyping as subjective cri‐ teria. They find that some of the categories are not at all applicable. One main finding is that salience cannot be reduced to single factors but rather is a complex process, e.g. salience does not seem to have an influence on strong accent fea‐ tures in their case study but only on features they classify as weak forms. Therefore, Auer, Barden and Grosskopf (1998: 184) come to the conclusion that “subjective and objective parameters in determining salience are therefore not mutually predictive.” Based on the results of the case study, Auer, Barden and Grosskopf (1998: 168) criticise Schirmunski’s (1928/ 29) as well as Trudgill’s (1986) concept in that salience in the two processes of levelling and diffusion are not seen separately. Labov (1972a) categorises linguistic features according to their awareness in a community, i.e. indicators, markers and stereotypes. For example, in the socio‐ linguistic interviews Trudgill conducted in Norwich he notes that he adapted some linguistic features of some of the interviewees. While Trudgill (1986: 7-11) accommodates his use of T-glottaling to the interlocutor, his use of [aː] is fairly constant and does not conform to the use of this feature by the interlocutor. 2.1 Dialect contact 11 2 Auer (2007) distinguishes between positive and negative dialect accommodation. Trudgill suggests that T-glottaling must be what Labov defines as a marker while [aː] is rather an indicator in Norwich English. Linguistic indicators show only limited style shifting while markers are subject to it. The comparison of indi‐ cators and markers on an awareness level shows that speakers are usually not aware of variation of indicators while markers are sometimes commented on, and style shifting of markers points to awareness of the variation. Accommodation has to be seen as umbrella term for two processes: Conver‐ gence seems to be the expected result of accommodation. However, linguistic divergence, though less well reported, is a possible outcome as well. Hickey (2000) suggests the concept of dissociation in opposition to accommodation: The notion of dissociation is diametrically opposed to accommodation, the approxi‐ mation of individuals to the speech of their interlocutors. The latter is taken to be - and have been - a powerful force in dialect differentiation as pointed out by socio‐ linguists repeatedly, above all by Peter Trudgill (Hickey 2000: 303). Figure 2.1 and Figure 2.2 present the differences in concepts between the short-term social psychological model by Giles and Powesland and the dialect contact model by Hickey. Giles and Powesland's concept of accommodation is divided into convergence and divergence. 2 For Hickey on the other hand, dialect contact situations are split up into accommodation and dissociation instances. However, linguists tend to use the terms divergence and dissociation synony‐ mously, e.g. Watts (2005: 9). Giles and Powesland (1975) promote a model where a short-term contact sit‐ uation is the point of departure. The contact scenario leads to the socio-psy‐ chological process of accommodation which can take the form of convergence or divergence towards a speaker. Trudgill (1986) applies this model to short-term dialect contact scenarios and hypothesises that convergence and divergence are also the outcomes when two speakers of different dialects come in contact. 2 Theoretical framework 12 Figure 2.1: Accommodation as described by Giles and Powesland (1975; cited in Trudgill 1986) Hickey (2000) on the other hand criticises the fact that the term accommodation only describes the process where speakers attune to their interlocutors but not the process where speakers distance themselves linguistically from their inter‐ locutors. This process is also not very well recorded in linguistic studies; though there are instances of this behaviour observed in ethnographic studies, e.g. Jones (2012: 198, 200-203). In his model, Hickey (2000) describes accommodation as the positive outcome of a dialect contact situation, with dialect convergence and dissociation as the negative effect, i.e. dialect dissociation. Figure 2.2: Accommodation process as described by Hickey (2000) 2.1 Dialect contact 13 3 Markedness is a concept which is often confused with salience. Marked features are features that are typologically infrequent, i.e. a linguistically objective category while salience is relative to the speaker group. Trudgill (1986: 98) defines markedness as “un‐ usual or in a minority.” A feature that is salient to one group of speakers does not have to be salient to another group. This fact has consequences for the language change process. Both models can be justified. While Giles and Powesland (1975) focus on the social psychological side of accommodation, Hickey (2000) concentrates on the (long-term) linguistic outcome. However, an important point which has to be made is that while accommodation towards an interlocutor is foremost an un‐ conscious process, dissociation or divergence in contact situations is very likely a conscious decision that a speaker makes. Yet this assumption needs to be tested. Researchers also have to trial whether accommodation has repercussions on long-term dialect contact outcomes. Divergence patterns are widely recognised, i.e. young people in a speech community try to dissociate from older speakers, MC speakers dissociate from WC speakers and vice versa. For instance, Watt (2002) suggests that young speakers of Newcastle English do not want to be identified with the old “cloth cap and clogs” image of the city anymore and therefore turn to a more general Northern feature which he interprets as dialect levelling. In a similar vein, Hickey (2005) explains the recent changes in Dublin English. But not only is dissociation likely due to age but also due to social practice (cf. Moore 2010). Geographical dissociation is observable in the north-east of England, where Burbano-Elizondo (2006) and Llamas (2007) have shown that some groups do not want to be associated with Newcastle and therefore linguistically dissociate from its linguistic norms. 2.1.2 Levelling One possible macro-linguistic outcome of dialect contact is levelling. In recent years, this geolinguistic process is a ubiquitous finding in European languages (cf. Foulkes and Docherty 1999b; Britain 2002; Kerswill 2003 for Britain; Boughton 2005 for France; Hinskens 1996 for the Netherlands). Even though many researchers had already stated that dialect features are lost, Trudgill (1986: 98) proposes a model of levelling as a process in which contact between speakers with different dialectal backgrounds leads to the levelling of local dialects. He defines levelling as “the reduction or attrition of marked  3 variants” (emphasis in original). 2 Theoretical framework 14 Kerswill (2003: 224) unravels the terminological ambiguity of the term level‐ ling in that he distinguishes between the two related concepts regional dialect levelling (= supralocalisation/ diffusion) and levelling. He argues that the former concept is an outcome of geographical and social processes. The latter term should be used in the sense that Trudgill (see above) referred to it: the process of delimiting the linguistic variables of a variety. Watt (2002) provides evidence that the local diphthongal variants of FAC E and G OAT in Newcastle English are undergoing a levelling process and are replaced by the pan-northern monophthongal forms which young speakers associate with ‘modern northerners.’ They do so by adopting features which are perceived as non-local. Foulkes and Docherty (1999a: 14) point out that “it seems to be important, too, that the incoming features do not signal any other particularly well-defined variety, because of the potential signalling of disloyalty to local norms.” Even though one can detect convergence of varieties in addition to the spread of innovative forms nationwide, these changes are locally determined and have to be analysed under this premise. “The specific attributes of any given locality are paramount in the explanation of the emergence of a supralocal variety” (Watts 2005: 19). On the surface geolinguistic processes might be accountable for variation within a variety but underlying patterns might indeed be the driving force for the variation. Thus, a detailed analysis of a change is necessary in order to detect and explain linguistic patterns. 2.1.3 Geographical diffusion Another geolinguistic process that recently has attracted a lot of attention from scholars is geographical diffusion. In recent years, many studies in the dialect contact framework have been dealing with geographical diffusion of linguistic features. Boberg (2000: 1) describes geolinguistic diffusion as “the process by which linguistic changes spread geographically from one dialect or language to another” while Kerswill (2003: 223), defines it as “features [that] spread out from a populous and economically and culturally dominant centre” which entails a hierarchy. In particular consonants in British English are “torchbearers of geo‐ graphical diffusion” (Kerswill 2003: 231), i.e. mainly consonants are diffusing across the country in a fast manner. The diffusion of features is enhanced by loose social networks when a change progresses through a community. Mobility is the main reason for close-knit networks to break up into loose social networks. Close-knit networks are more traditional in their language choice than loose networks which often are the 2.1 Dialect contact 15 engines of diffusion. Individuals in loose networks are in contact with many people often on an infrequent basis, and often from outside a community which leads to various dialect contact situations. Therefore, they are usually the ones who adopt changes easier and faster than people in a closed circle of friends and family who are usually restricted in their contacts to small areas (cf. Milroy 1987: 202-203, 213). Inter-speaker communication is thus the core issue for supralocal changes. Britain (2010b) emphasises the importance of physical and social space for the diffusion of linguistic features. The habits and daily routines of individuals have changed dramatically over the course of the 20 th century (cf. Watts 2005; Britain 2010b). Mobility, commuting, out of town shopping and attending university away from home are all social practices that have developed and increased only in the last sixty years or so (cf. Beal 2010). It is very likely that these factors are influencing the diffusion of linguistic features which is observed in Britain. Spatial practice (Britain 2010b) and mobility (Milroy 1987) are both interrelated factors which can have linguistic effects. Thus, spatial practice might be a social factor which influences dialectal variation. Wolfram and Schilling-Estes (2006: 154) emphasise that dialect diffusion can happen on a horizontal dimension (geographically) as well as on a vertical di‐ mension (socially). If a diffusional change arrives in a community, it is usually one social group that promotes the change before it extends to the use of other social groups. Thus, the use of linguistic features can be indexical for being part of a certain group. For example, the use of glottal stops used to be indexical for Cockney speakers, i.e. WC speakers living in London’s East End. This enregis‐ terment has decreased in London recently due to the spread of this feature to other vertical dimensions. Thus, today T-glottaling is also used by middle class speakers in London (cf. Altendorf and Watt 2008). This linguistic feature has also spread horizontally so that the use of T-glottaling is found in other areas around Britain. Chambers and Trudgill (1998: 167) identify four ways in which sound change can diffuse within a variety. 1. Sociolinguistic diffusion: social group to social group 2. Lexical diffusion: word to word 3. Linguistic diffusion: linguistic environment to linguistic environment 4. Spatial diffusion: place to place These kinds of diffusion usually do not happen in isolation but are inter‐ twined. In this particular study I will analyse the sociolinguistic, linguistic and spatial diffusion of selected features of Carlisle English. 2 Theoretical framework 16 Beal (2010: 73) states that “changes involving diffusion have an identifiable historical starting point and geographical trajectory: ‘new’ variants can be seen to have spread or to be spreading, displacing ‘old’ variants indigenous to the places affected.” For the consonantal changes observed in Britain in recent years, it is often assumed that some these changes started in the south-east and in particular in the East End of London (with the exception of T-glottaling which seems to have originated in two separate regional centres (cf. Beal 2007; Schleef 2013; Smith and Holmes-Elliott 2017). Diffusional changes do not spread in a uniform way. Britain (2009: 137f) gives an overview of different diffusional mechanisms. The diffusion of linguistic fea‐ tures in a wave or contagious manner describes a model of a spread of innovations from a central area, first to places in close proximity to the centre and then to places which are geographically further away. Boberg (2000) describes this dif‐ fusional pattern for (a) from the USA to Canada. On the U.S. side of the border, the innovation first reaches areas that are geographically closer, before it spreads to centres in Canada. Kerswill (2003: 234) claims that TH-fronting and other consonantal variants are diffusing in this manner in the south of England and then across the country. A second model of diffusion is the urban hierarchical or cascade model where an innovation reaches urban areas earlier than smaller, geographically closer areas. Labov (2003) claims that the Northern Cities Shift in the USA can be interpreted as a diffusional change according to the urban hierarchical model. He argues that this change was first observed in the largest cities Chicago, De‐ troit and Buffalo, but now involves a large number of smaller cities. Bailey et al. (1993: 361) argue that the gravity model, introduced by Trudgill (1974: 235), is a model to explain and predict hierarchical diffusion by calculating the interaction between two places on the basis of overall population and dis‐ tance. Even though diffusion is explained by the gravity model in sociolinguistic studies, e.g. West (2013), there are some issues. First and foremost, the model does not take any geographical boundaries such as hills, lakes or valleys into account which could prevent contact situations between speakers. In addition, Bailey et al. (1993: 361) point out that linguistic diffusion might not be a clear-cut business. They mention important factors which have to be taken into account when data is discussed on the ground of geolinguistic diffusion models. […] our research shows that the emergence of focal areas for linguistic diffusion is not solely a consequence of the size of the population of those areas. The demographic makeup of the metropolises, their location in particular dialect areas, and their prox‐ 2.1 Dialect contact 17 imity to innovations spreading from other areas all contribute to their emergence as focal areas. Wolfram and Schilling-Estes (2006: 154) provide examples where the gravity model fails due to geographical isolation, e.g. isolated parts of the Southern mountain ranges of Appalachia and some of the islands off the Atlantic coast such as Smith Island, Maryland. They point out that “this model rarely works out neatly or symmetrically” (Wolfram and Schilling-Estes 2006: 155). They refer to the work by Rogers (1995), who names five factors which (at least) influence the diffusion of cultural innovations, such as customs: “(1) the phenomenon itself, (2) communication networks, (3) distance, (4) time, and (5) social struc‐ ture.” Though Wolfram and Schilling-Estes point out that linguistic diffusional processes are quite different from other kinds of (cultural) diffusion, they argue that a wave diffusion model can hardly be reduced to distance and time (cf. Britain 2010a: 149-151; Britain 2002). The gravity model also fails to explain cases of resistance to diffusion. Watson (2006) reports that the use of glottal stops for (T) is not attested in Liverpool English. Merseyside seems to be resisting the change while the feature has in‐ deed diffused to other varieties in surrounding areas such as West Wirral (New‐ brook 1999). Watson (2006) argues that the reason for this resistance is the lan‐ guage internal lenition process with the variants slit-t and [h] for the (T) variable which stand in structural contradiction to T-glottaling, even though the use of glottal stops can be interpreted as a lenition process as well. Thus, not only do language external factors play a role for the diffusion of features, language-in‐ ternal factors do as well. The cultural hearth diffusion model explains a diffusion pattern which spreads from an urban centre to the surrounding area first before it spreads to other urban areas. For example, this pattern is attested for Australian English. Hor‐ varth and Horvarth (1997, 2001, 2002, cited in Britain 2010a: 148) show that L-vocalisation is attested in and around Adelaide first and then spreads to other regions of the country instead of diffusing from the central focal areas Sydney or Melbourne. A fourth model is the contra-hierarchical diffusion. In this model, innovations spread against the urban hierarchy, i.e. they are observed in rural areas first and then spread to urban areas. This kind of diffusion is rarely found. Trudgill (1986: 47-49) gives an example from his East Anglian research where smoothing effects reduce triphthongs such as in tower to monophthongs in the more rural Northern part of the East Anglian region and then spread to the larger towns. Overall, the diffusion of linguistic features is not restricted to one type but it is possible to find different patterns for one community (Bailey et al. 1993). They 2 Theoretical framework 18 4 Despite Wright (1979) comment that (L) is clear in all instances in Cumbria, L-vocali‐ sation seems to be a long-established feature in Carlisle English, in particular after high back vowels. Further investigation into the variety is needed in the future. also point out that the social meaning of a feature is crucial for the spread of it, an issue which is also emphasised by Britain (2010b). The question how and why changes diffuse has been of particular interest for some years now since a range of features have been identified as spreading across Britain. A large number of studies report the diffusion of these non-standard features to various communities around the country (Williams and Kerswill 1999; Watt and Milroy 1999; Foulkes and Docherty 1999a; Trudgill 1999a; Llamas 2007; Clark and Trousdale 2009; Jansen 2014 among others). Kerswill (2003) discusses a row of consonantal changes which seem to be spreading across Britain. He mentions: • [ʔ] for / t/ in intervocalic and word-final position as in butter and but (see §5.2 for further discussion). • [ʋ] for / r/ in place of [ɹ] in non-postvocalic / r/ position (see §5.3 for further discussion). • The merger of / θ/ and / f/ as [f] in word-initial position and the related merger of / ð/ and / v/ as [v] in word-medial and word-final position (see §5.4 for further discussion). • The vocalisation of / l/ in postvocalic position. 4 At the time, Kerswill (2003: 231) suggests that only consonantal variables are diffusing: “there appear to be no reports of vowel changes spreading throughout the whole country: local ‘solutions’ are the order of the day.” However, recently the fronting of G OO S E has been observed as linguistic change in many varieties in England (e.g. Flynn 2012 in Nottingham and Baranowski 2017 in Manchester). Not only the G OO S E vowel, also the G OAT vowel seems to be subject to suprare‐ gional change. The spread of the features has raised issues concerning standard language. A common assumption in the early 20 th century was that lost regional and local features would be replaced by standard features. This assumption has been challenged repeatedly. The process of levelling, where marked or minor features are eradicated from a variety, is indeed closely related to the standardisation process. Diffusion of linguistic features, on the other hand, can increase the variability within a variety. Thus, one can assume that long-term accommoda‐ tion and the resulting diffusion and levelling of linguistic features are focal topics in the area of geolinguistic processes. However, linguists still do not understand these phenomena in their entirety. It is not at all clear which features are can‐ 2.1 Dialect contact 19 didates for supralocal change and the extent of contact which must take place so that features spread. These are issues that have yet to be researched. 2.2 Regional identity Above, geolinguistic processes that potentially lead to language change as set out by Trudgill (1986) have been discussed. The section on accommodation clearly shows that these processes must have their beginning at the individual speaker level. However, on the macro level the discussion on diffusion is not complete without considering the social motivation within a community for change triggered by identity constructions. Torgersen and Kerswill (2004: 25-6) emphasise that internal and external as well as extra-linguistic factors have to be taken into account for the analysis of changes within a community that are observed across a larger geographical area. These extra-linguistic factors lead to “locally specific outcomes” (Watts 2005: 26). For Britain (2010b), these extra-linguistic factors are part of the spatial practice of the community. Nev‐ ertheless, they all agree that factors which lie outside of the linguistic system of a community can and do influence linguistic choices. It is important in which (physical and emotional) direction people orient themselves as a community. This extra-linguistic behaviour should be taken into when investigating the lin‐ guistic behaviour of speaker groups. However, spatial practice is often only partially responsible for variation and change processes. In particular in the north of England, several studies have shown that the attitude of speakers can have an influence on linguistic change (cf. Watt 2002; Burbano-Elizondo 2006; Llamas 2007). Britain (2010b: 203) calls for attention to the social practice of the speakers. Positive and negative associations with other regions can have linguistic repercussions, e.g. Llamas (2007) sketches the his‐ torical development of the city of Middlesbrough with various cases of geo‐ graphical reorganisation which is reflected in political, social and perceptual views of the speakers. Nevertheless, these social practices have repercussions on the linguistic choices of the speakers, repercussions which sociolinguistic studies are trying to detect. 2.3 Dialects in contact: The Carlisle setting The discussion in this chapter has provided an overview of the mechanisms of dialect contact situations that are relevant in various speech communities. The 2 Theoretical framework 20 present study is based on the dialect contact framework. Britain (2010b) argues that the increase of mobilisation has led to the increased contact between people with different dialects. Hence, he claims that this rise in contact situations has led to an advanced stage of levelling in Britain. While his argumentation is true for most contact situations which evolved due to better transportation or mo‐ bility across long distances, Carlisle is a microcosm in its own right. Increased mobility certainly plays an important role. People frequently commute to Carlisle from Scotland but also from West Cumbria and Northumberland and other surrounding areas, all of which have quite distinct dialects. Indeed, the geographical position of Carlisle could also have an influence on language choice in Carlisle. Carlisle is also the city in England which is furthest away from London and the south-east, the area where many of the features subject to consonantal dif‐ fusion seem to origin. One aim will be to research diffusional patterns in Carlisle English. Therefore, the description and analysis of features which are diffusing to the city or which are varying between speaker groups can provide us with information about the distribution of these diffusional features as well as soci‐ olinguistic information about the social make-up of the city. Recently two major projects have involved Carlisle as an object of research. Montgomery (2006) takes a perceptual approach to geographical variation with one of the fieldwork sites being Carlisle. He finds that people in Carlisle recog‐ nise their accent when it was played to them but when they were asked to place the accent on a map they placed it as far away as possible from the actual setting of Carlisle. He argues that because the people do not like the accent they react this way. A second study which focuses on the dialect of Carlisle English is conducted by Llamas et al. (Accent and Identity on the Scottish-English Border (AISEB) 2008-2011). They are concerned with ‘border identities.’ In their project, speech in Carlisle and three other towns along the Scottish-English border are analysed in terms of linguistically converging and diverging trends between border towns. For this project, data was collected in four towns in close proximity to the Scottish-English border: Greta (Scotland) and Carlisle (England) in the west and Eyemouth (Scotland) and Berwick-upon-Tweed (England) in the east. This study sets out to investigate language variation and change in the Carlisle dialect. As has been noted above, Northern England has received a lot of linguistic attention. Nonetheless, Carlisle English and the far north-west have so far been largely neglected. In particular the Tyne and Tees area, Liverpool and Sheffield (cf. Beal and Corrigan 2002; Watson 2006; Llamas 2007; Beal, Bur‐ bano-Elizondo and Llamas 2012), places that are frequently embedded in larger 2.3 Dialects in contact: The Carlisle setting 21 urban conurbations, are regularly the subject of linguistic studies. However, until recently little was known about the Carlisle variety of English, a city without an immediate urban neighbourhood. Description work on major ac‐ cents and dialects such as Wells (1982: 350f) hardly discuss or offer examples for a Cumbrian or Carlisle accent. For the Carlisle dialect in particular only one source of data was available up until recently. Peter Wright’s (1978) description of the Carlisle dialect was the only attempt to approach this urban area linguistically. This short study is not systematic and it is written in a rather popular style. He describes the features which are salient to him using semi-phonetic spelling which leaves room for interpretation. He claims that “although Cumbrian towns and villages are sin‐ gularly lucky in retaining so well many aspects of their traditional vernacular, Carlisle should never be forgotten” (Wright 1978: 7). With this statement he illustrates the dilemma of the Carlisle dialect. He hints at a handful of studies that have been conducted in the far north-west, especially in the former county of Westmoreland. For example, Reany (1927) deals with grammatical aspects of the Penrith dialect (Penrith is 36 km south of Carlisle), Brilioth (1913) focuses on the grammar of Lorton in Cumberland (Lorton is 55 km south-west of Carlisle) and Rydland (1978) investigates the dialect use of south-west Cumbria. However, no one has conducted a systematic study of the Carlisle dialect. Only lately have linguists started to comment on this desideratum. Beal (2008) argues that there are certain areas in Northern England which have not been covered in sociolinguistic and dialectological terms and Carlisle is one of them. Wales (2008: 172) comes to the same conclusion: “Manchester, the ‘capital’ of the north in the nineteenth century, Leeds and Bradford with their significant minority group populations, Lancaster and Carlisle, […], all await investigation.” Since then a number of studies on the Carlisle dialect have been conducted, e.g. Jansen (2015a); Jansen (2015b); Jansen (2017a); Jansen (2017b). 2 Theoretical framework 22 3 Carlisle and its geographic, historic and sociolinguistic background The aim of this chapter is to provide information about the geographic and demographic background as well as the historic development of Carlisle as an urban centre in the north of England in order to ascertain important develop‐ ments which have had and still have repercussions on the linguistic develop‐ ment of Carlisle English. 3.1 Geography The county of Cumbria was created in 1974 after the Local Government Act (1972) by combining the historical counties Cumberland, Westmorland and Fur‐ ness, formerly a part of Lancashire. The county is divided into six districts: Allerdale, Barrow-in-Furness, City of Carlisle, Copeland, Eden und South Lakeland. After North Yorkshire and Lincolnshire, the county is the third largest non-metropolitan county in England. Neighbouring counties of Cumbria are Northumberland, County Durham, North Yorkshire, Lancashire and the Scottish council areas of Dumfries, Galloway and Scottish Borders. To the west there is the Irish Sea with the Solway Firth. The Lake District in the south of the county is a tourist attraction of which Carlisle as the largest and only city in Cumbria can only partially benefit. Carlisle is not a typical holiday destination but more of a stop off city. Carlisle is also known as the Border City. This title reflects its proximity to the Scottish border which is only 16km away. Because of its geographical key position in Britain, Carlisle has always been important as a military defence point. While Newcastle was founded in the east to secure the area, Carlisle was seen as the equivalent in the west of the country. Map 3.1. shows Carlisle’s geographical location in England. London is about 500 km away which makes Carlisle the English city which is furthest away from London. This is an impor‐ tant point considering the diffusion of linguistic features which are associated with London English. The city’s surrounding area is dominated by agriculture. The closest urban area is Tyneside, some 90km away to the east, accessible by rail and car. South-west of the city, about 50km from Carlisle, one finds the Lake District with its rugged landscape. To the east are the Pennines, which stretch across the country. They are not very high in the north-east, making travel to Newcastle relatively easy. The Borderlands, also known just as the Borders, as well as the area west of Carlisle out to the Irish Sea (West Cumbria) are sparsely populated areas. The setting of the Carlisle and its surroundings makes it interesting in terms of identity construction but also in terms of geolinguistic spread, i.e. lev‐ elling and diffusion of features. Figure 3.1: The geographical position of Carlisle in Britain (d-maps.com) Figure 3.1 provide a map of parts of the UK and Cumbria. Carlisle is connected with other regions by road and rail. The only motorway in Cumbria is the M6 which runs through the county on a north-south axis connecting Cumbria with Edinburgh/ Glasgow to the north and Lancaster, Merseyside and Greater Man‐ chester to the south. The city of Carlisle is found in close proximity to this motorway. In addition, the A595 and A596 coming from Whitehaven and Work‐ ington end in Carlisle and the A69 connects Carlisle with the north-east. Thus, Carlisle has good road infrastructure. 3.2 Demographics According to the 2011 census, Carlisle has a population of 104,500. The British Census data from 1831 to 2001 shows a gradual growth from around 40,000 in 3 Carlisle and its geographic, historic and sociolinguistic background 24 1831 to around 100,000 in 1961. After this, the number of people levels at about 100,000 (www.visionofbritain.org.uk). The data in itself shows that there are no large groups suddenly swamping the city. Carlisle is a city with a long history which has not seen any drastic population changes in the last fifty years. How‐ ever, even though the data paints a picture of gradual and at times stagnating population growth, there has of course been some movement. Out of the 40 participants in this study, 34 provide some information on the background of their parents. Either one or both parents of 26 people in the sample were not from Carlisle originally. This was also my general observation when I was trying to find speakers in Carlisle for my sample. In many cases, they were not origi‐ nally from Carlisle but from surrounding areas and it was also not uncommon to meet people from further away, e.g. Malta, Australia, Middlesex and London. I also asked the people in which areas of the city they had lived. Most of them had moved within the city. Thus, one can assume that close-knit communities did not exist or were weakened by this inner-city movement. The influx of people to Carlisle is not the whole story. Shepherd (2011) reports quite a large number of emigrants from Cumberland between 1815 and 1914. While she shows that many of the people who left the county were from more rural areas, there were also emigrants leaving Carlisle. Hence, while there was an influx of people from Scotland, Ireland and West Cumbria to Carlisle in the 19 th century, at the same time people were also leaving the city. Towns found overseas with names such as Carlisle in Cumberland County, Pennsylvania and Penrith in Australia are evidence for these emigration tendencies. In terms of ethnic groups, the vast majority of the population (97.8%) was ‘White British’, a classification used in the 2001 census. 85.7% of the United Kingdom was classified as ‘white’. Carlisle is therefore one of the whitest places in England. However, since 2001 the city has seen an influx of various (white) ethnic groups. It is estimated that a population of around 1,600 Polish workers and 500 Portuguese workers live in Carlisle (Carlisle City Council Equality Im‐ pact Assessment: 6). Thus, ethnic minorities play a minor role in the population make-up of the city. According to the 2001 census, 35.7% of the population belonged to the Middle Class (nationally: 48%). The comparatively low percentage of Middle Class people in Carlisle is likely to have two reasons. 1) The number of pupils who leave school without graduating is comparably high: 32% in Carlisle vs. 21% in Cumbria (Huntington and McKerr 2008: 2). 2) There is a lack of jobs for more highly qualified people. Even though Carlisle now has a university, it only offers a limited choice of degree courses. For this reason, many young people leave 3.2 Demographics 25 Carlisle to study elsewhere never to return because finding employment in Carlisle is difficult, especially in fields such as media, marketing, and academia. The two main employment sectors are and have been factory work (manu‐ facturing) and services. Agriculture does not play an important role anymore, at least not in the city. 3.3 History The history of the city dates back to the Romans. They installed a military base, building a fort in the Carlisle area around 80 A.D. which they called Luguvalium (Smith 1984: 8). From this fort, the Romans unsuccessfully tried to conquer Scotland. When they realised that they could not defeat the Scots, one of the most important Roman sites, Hadrian’s Wall was built in 122 A.D “to keep out the barbarians” (http: / / www.hadrians-wall.info/ ). The power vacuum left by the Romans was filled with bloody conflict between different Celtic tribes. This period of instability finds its climax in the raids of the Danes (Smith 1984). Up until these Viking raids the area was Celtic-speaking. Only after the Viking invasion did English become the language spoken in this area (Smith 1984). Carlisle only slowly recovered from the Scandinavian incursions. When Wil‐ liam II came to Carlisle in 1092 his goal was to restore the town and build a castle in order to secure the frontier between England and Scotland. During the fol‐ lowing centuries, Carlisle was often under siege and was torn between England and Scotland. Because of its ideal position as a military stronghold, both coun‐ tries made demands on it. Towill (1996: 137) summarises the situation of Carlisle in the middle of the 18 th century, after the last siege, as follows: Carlisle in 1745 was a beleaguered fortified city, a city with a self-centred unspecialised economy with citizens providing for their own needs and making their own clothes, and an isolated city with minimal communication with the rest of the country. From the mid-18 th century onwards the situation changed and Carlisle became less isolated, gaining importance as a communication centre in particular for the military. To make travel between Carlisle and Newcastle easier, a road was built along Hadrian’s Wall. In 1759 a visitor described Carlisle as “a small deserted dirty city; poorly built, and poorly inhabited. The Cathedral is miserably ragged and dirty inside and out” (quoted in Smith 1984: 85). At this time about 2,000 people lived in the town, but by 1763 this number had increased to about 4,150. The years up to the turn 3 Carlisle and its geographic, historic and sociolinguistic background 26 of the century saw a vast number of arrivals from Cumberland, Scotland and Ireland. In 1801, the city had as many as 33,000 inhabitants and this number doubles by 1901, reaching 75,000 (www.visionofbritain.co.org). Already in the year 1770, which many see as the beginning of the industrial revolution, Carlisle played a major role in the textile industry in the north-west of England. Many hand-loom weavers came to the city during this period. Up to around 1815, workers in the textile industry earned high wages, but wages started to decline. This, however, did not stop new people from moving to Carlisle to take up the trade, which in turn had a negative effect on wage de‐ velopment and increased urban density. Eventually, the workers were replaced by power looms, widening the social gap between the rich and the poor (Towill 1996: 53-58). Towill (1996: 60) gives a short impression of the social conditions in 1819: “one half of the population of the city live in dire poverty and another quarter only survived with difficulty.” Social conditions were further exacer‐ bated by bad housing, disastrous sanitary conditions in the inner-city area called The Lanes and overcrowding. Due to the population increase the city walls were demolished in the 1820s. The Lanes was one of the most important areas in the city centre, which had been built in Medieval times. The area was known for its small houses and narrow lanes. In the early 1980s the area was demolished after a house had collapsed shortly before HRH Queen Elizabeth was to arrive in Carlisle in 1979. As mentioned above, the beginning of the 19 th century Carlisle saw a strong influx of Scottish and Irish immigrants. Even before the Irish Potato Famine (1845-1851) many Irish workers arrived at the west coast of Cumberland in towns such as Maryport and Whitehaven only to move on to Carlisle. Their living situation is comparable to that of Irish immigrants in Sheffield (Beal and Corrigan 2009) who lived only in a particular part of town, separated from the rest of the community (Armstrong 2003). Seven railway companies operated in the city (McCarthy 1993: 87), making Carlisle one of Britain’s most important railway junctions (Towill 1991) until the mid-20 th century. In 1838, a railway connection between Carlisle and New‐ castle was opened and in 1845 a further connection to Maryport in West Cum‐ bria. In addition, in 1881 the Border Regiment was formed and soldiers from around the country came to Carlisle. Many soldiers and railway workers even‐ tually settled for good in Carlisle. Unfortunately, census data does not provide detailed information on where exactly these people came from originally. At the end of the 19 th century, textile, railway, engineering and metal were the major industries in Carlisle. All these industries saw decline during the 20 th century (Towill 1991: 123). In the latter half of the 20 th century, many traditional 3.3 History 27 companies such as Carr’s (Carr’s Biscuits) and Metal Box were taken over by larger, international companies and companies such as Pirelli and Nestlé even settled in Carlisle (Towill 1991: 124). With its 34% working class population (26% nationally, Census 2001), Carlisle can today be seen as a working class town. However, many factories have closed down over the years and Carlisle has in‐ creasingly become a town with a strong service industry, attracting many com‐ muters from surrounding areas. In fact, the trend of commuting from south-west Scotland and from other parts of Cumbria to Carlisle has long been recorded (Coombes 1995: 47). As a response to population growth in Carlisle in the 19 th and early 20 th century, a housing estate called Raffles was set up in the 1920s and 1930s. Parts of the estate were then demolished in the 1990s and families were rehoused, potentially splitting up close-knit communities. It is most likely that moving families from Raffles and the Lanes area to other parts of the city has had an effect on linguistic choices. Montgomery (2006) conducted a study researching how Carlisle and several other Northern cities are perceived by locals as well as outsiders. He says about Carlisle that “this lack of prominence produced interesting maps, with blank spaces indicating the perceptual ‘black hole’ in the north-west of the country” (Montgomery 2006: 221). Burbano-Elizondo (2006: 114) states that “for outsiders there does not seem to be anything else in the North-East apart from Newcastle and Geordie.” This is also true for the Carlisle region. Newcastle and Geordie are the predominantly perceived cities and varieties in the far north of England. In fact, many speakers in the interviews for this study mentioned that their variety had been mistaken for Geordie and/ or Scottish when talking to strangers. One reason for this is that economically Carlisle plays only a minor role in Northern England. Another reason may be that Carlisle has in the past not been very well represented in the media. Many speakers in the interviews mentioned that Cumbria no longer has its own TV station and that ‘Border TV’ covers the north-east far more than the far north-west. National weather forecasts often do not even include Carlisle while covering the much smaller town of Kendal which is situated in the Lake District. In 2005 (and again in 2015), Carlisle was struck by massive floods. On both occasions the city received national media coverage. On a more positive note, in 2011 Carlisle hosted the BBC1 Big Weekend. The local newspaper Cumberland News (9 September 2011) rated the event as a complete success: “The report […] adds that although the ‘feelgood factor’ cannot be quantified, it is regarded as one of the key impacts of the Big Weekend and is responsible for an ‘increased confidence’ in the city.” The report further states: “The Big Weekend has also 3 Carlisle and its geographic, historic and sociolinguistic background 28 ‘put Carlisle on the map’ for many people who would not have visited the area before or may not have even known where the city was.” 3.4 Carlisle in the northern English context Of all cities in northern England, Newcastle has the greatest influence on Carlisle. For people from Carlisle, Newcastle has been a centre of attention for a long time. Already in 1838, a daily railway connection was established between the two cities (McCarthy 1993: 86) and Carlisle people are still travelling to Newcastle quite frequently. Even though some people commute to Newcastle for work, the majority of Carlisle people go there for shopping and pleasure, e.g. for concerts and theatre plays etc. The sort of (football team) rivalry found in the different urban areas of the north-east, e.g. Sunderland vs. Newcastle, or Middlesbrough vs. Newcastle, does not exist between Carlisle and Newcastle. Watt (2002) explains that historically Northumberland and Scotland were closely connected and that this connection is still felt in the city. Even though Carlisle shares history with Scotland, people in Carlisle, however, do not really feel connected with Scotland. Historically, Hadrian’s Wall was a cut-off point. This is reflected in the very limited information overlap between the Carlisle archive and archives across the border (local archivist p.c.). McCarthy (1993: 99) makes this point very clear when he talks about Carlisle identity: In many respects the history of Carlisle compares well with many other towns. First innovation, then stability, and finally, rapid change in the last two centuries have characterized it. What marks Carlisle out as different from many other places is its geographical isolation and its frontier position. The people of Carlisle have always identified closely with their city and its surroundings. Perhaps its remoteness and its border location are partly responsible for the conservative element in local society, but that element is deeply rooted in history. As one writer put it, ‘Cumbrians, when border warfare was over and masstrooping days had passed away, still felt that they were not as other men are.’ Thus, Carlisle people define themselves based on the geographical position of their city. In terms of identity, Carlisle does not seem to have emancipated itself from its Cumbrian roots. When asked about the Carlisle dialect, people inter‐ viewed for this study often talked about the traditional Cumbrian dialect. More‐ over, a regional nickname such as ‘Geordie’, ‘Mackem’, ‘Scouse’ or ‘Cockney’ has not development yet. This may show that the inhabitants of Carlisle do not yet see Carlisle English as a variety in its own right, different from the varieties 3.4 Carlisle in the northern English context 29 spoken in the areas surrounding the city. In fact, a Google search (11.10.2011) did not give any results for ‘Carlisle identity’, while ‘Cumbrian identity’ did give at least some results while there is even a research project devoted to ‘Geordie Identity’ now. 3 Carlisle and its geographic, historic and sociolinguistic background 30 4 Methodology The data presented in this book stem from recordings of sociolinguistic inter‐ views with people from Carlisle. My fieldwork in the community took place in two major stages in February/ March 2007 and February/ March 2008 with the addition of three individual recordings in September 2009 and two further re‐ cordings in September 2010. In all, I made eighty recordings, recording a total of 109 speakers. The recordings lasted from around 25 to 90 minutes. Participants were recorded individually, in pairs and in one case in a group of three. In the following I discuss the sampling decisions, methods and interview procedure. In the first section I discuss the social variables age, sex and social class as external motivations of change (cf. Labov 2001). 4.1 Age as external factor The age of the participants of this study ranges from 15-99 and the adjusted sample for this study ranges from 22-78. The more traditional approach is to categorise speakers in age cohorts, usually according to life stages (e.g. Trudgill 1974; Hall 2008). Fewer, but not less convincing studies (e.g. Hall-Lew 2009 and to a certain extent Baranowski 2007) use age as a consecutive variable. In this study, the vowel and consonant variables are statistically analysed taking speaker age into account as a continuous predictor. The figures for the conso‐ nantal variables are still presented with age categories in order to make the results more accessible. The age cohorts are 22-39 (young), 40-59 (middle), 60-78 (old). Age is not only important on the individual level but also on the speech com‐ munity level. In sociolinguistics, age is represented in real-time and appa‐ rent-time studies as tool to investigate language change. Labov (1963) introduced a set of methods in order to tackle the problem of analysing language change in his Martha’s Vineyard study and later in his New York study (Labov 2006). His approach of combining diachronic and synchronic approaches in order to study language change is a paradigm shift away from structuralism where language change is believed to only be observed diachronically but not synchronically. Sankoff (2006: 113) describes how sociolinguists benefit from the appa‐ rent-time approach: The most important implication […] is that apparent time is a truly powerful concept in locating the presence of change. […] a researcher who locates a gradient age dis‐ tribution in a new community under study is virtually assured of having identified change, whether or not age grading is also involved. Di Paolo and Yaeger-Dror (2010: 9) stress that “the speakers’ age at the time of the interview is primarily important because it marks their life-stage […], and their progression through the linguistic market” (their emphasis). Thus, the speaker’s behaviour and attitude towards socially loaded variants can change over the course of a life, i.e. the importance a speaker gives to the use of certain variants varies. 4.2 Sex as external factor In this study, the social category sex is based on the biological categories male and female. Sex differences are particularly interesting when it comes to the question of who leads linguistic change. Labov states that “in the majority of linguistic changes, women use a higher frequency of the incoming forms than men” (Labov 1990: 206). As Chambers (2009: 256) points out, this statement is somewhat vague because most changes are towards standardised forms. There‐ fore, Labov (2001: 293) refines his claim in the Sex Paradox: “Women conform more closely than men to sociolinguistic norms that are overtly prescribed, but conform less than men when they are not.” In general, women tend to use more standard forms than men because women’s perception of non-standard use of language is sharpened and women, in the majority of cases, lead language change (Wodak and Benke 2017: 138). However, male speakers lead change to‐ wards non-standard variants. For example, Trudgill (1972) shows that WC male speakers lead the change towards the use of [ɪn] instead of using the more standard variant [ɪŋ]. Likewise, Williams and Kerswill (1999: 160) find that WC (Working Class) boys in Reading and Hull are leading the change towards [ʔ] for (T). Even though Chambers (2009: 105) states that “very few biological differences between males and females have an effect on language”, in instrumental/ acoustic phonetics physiological differences between men and women have to be mentioned. Male speakers tend to have a larger larynx than female speakers. The size of the larynx influences the size of the vocal cords, i.e. males also have longer vocal cords which in return means that females have higher Hertz fre‐ quencies when speaking. In general, the aim of sociophonetic studies is to com‐ pare speaker groups and thus, normalisation procedures have to be applied to 4 Methodology 32 1 Income as a factor is quite problematic and usually off-limits when linguists approach people. Researchers have tried to overcome this problem by assigning people to certain areas within the cities or just designing the social class factor as a combination of ed‐ ucation and occupational status. the data in order to extract physical sex differences (cf. Chambers 2009: 106f) by acoustic means. 4.3 Social class as external factor The classification of interviewees into groups based on social class is not a straight forward task and has been discussed widely (cf. Mallinson and Dods‐ worth 2010). Kerswill (2009: 361) postulates that “there is no ‘natural’ way of defining social class.” While in most westernized countries social class differ‐ ences are no longer very strong, “in Britain, a survey found that 36 per cent of adults considered themselves ‘middle class’, while 46 per cent viewed themselves as ‘working class’, reflecting a relatively polarized view” (Aryle 1994: 4, quoting Reid 1989, quoted in Kerswill 2009: 363f). Social class is, however, not always an obvious criterion to distinguish. Socio-economic indices are a helpful tool as for example introduced by Labov (2006) to determine the socio-economic status of a speaker. But this tool is not unproblematic. Chambers (2003: 49) talks of an “abstraction” of real life when factors such as income, 1 education and occupational status are combined. In the present study a two-component composite index modelled after Labov (2001: 61) and Hall (2008: 64ff) is applied to classify the socio-economic class of the speakers. The scores of the educational and the occupational indices were added up which then resulted in the socio-economic score of each speaker. The classification of the speakers’ education level was not as straightforward as expected. The education system in England saw many changes during the 20th century. Therefore, the questionnaire I passed out could not contain specific school-leaving certificates. Instead, I opted for more general terms to describe their level of education. Table 4.1 provides the categories used in the question‐ naire. Points 1 No leaving certificate 2 Standard Grade (equivalent to GCSE/ O-level) 4.3 Social class as external factor 33 3 Advanced Higher Grade (equivalent to A-level) 4 Tertiary Education Table 4.1: Education levels In addition to these education levels, the occupation of the speakers was rated. Table 4.2 presents the occupational types introduced by Labov (2001: 61). Other indices such as The National Statistics Socio-economic Classification published by the UK Office for National Statistics in 2001 were considered as an alternative to Labov’s classification. However, his occupation types proved to work best for this study. Points Occupation type Example 1 Unemployed 2 Blue-collar - unskilled factory worker 3 Blue-collar - skilled painter and decorator 4 White-collar, lower level administrative assistant 5 White-collar, higher level speech therapist 6 Professional teacher Table 4.2: Occupation types In Table 4.2 I also give examples taken from my sample for each of the occupation types. Table 4.3 provides the socio-economic category distinction (SEC) adapted from Hall (2008). Unlike Labov (2001) and Hall (2008), in this study the speaker groups are not categorised into four SEC classes but only two social classes. I combined the original Lower Working Class group and the Upper Working Class group into the Working Class (WC) group; and combined the Upper Middle Class and Lower Middle Class into the Middle Class (MC) group. SEC points Socio-economic class 6-10 Middle Class 2-5 Working Class Table 4.3: SEC groups for this study (adapted from Hall 2008: 69) 4 Methodology 34 Labov (2006: 177ff) uses an index which is based on three categories: occupation, income and education. Trudgill (1972) on the other hand uses an index which includes type of housing, locality and father’s occupation. I did not ask the par‐ ticipants about their income, for fear of an awkward situation which would then potentially lead to the subject rejecting an interview. Housing is another cate‐ gory which is often applied. Not all interviews for this study took place at the people’s homes and therefore I could not always judge the socio-economic background of the interview partners. 4.4 Sampling methods and contacting speakers Speakers from Carlisle and surrounding villages which are not more than five miles away from Carlisle were included in the sample. Four of the speakers are from surrounding villages but all speakers went to school in Carlisle. One of the main challenges in a variation and change study is to collect a convincing sample. “The key to achieving a representative account of the language of a group of speakers is the avoidance of bias” (Milroy and Gordon 2008: 24). Re‐ searchers are at risk of compiling a biased sample when they choose speakers based on their language use. In other words, the researcher would choose speakers depending on what the researcher hopes to find. In order to avoid a rigid, often problem-causing sampling method, sociolin‐ guists often apply the stratified random sampling technique, which is also applied in this study. Milroy and Gordon (2008: 30) point out that it is the task of the researcher to identify possible social factor groups prior to the fieldwork in order to categorise speakers according to external factors. Tagliamonte (2006) pro‐ poses that “stratified random sampling requires extralinguistic justification for its selection criteria, whether sociological, demographic or otherwise.” There‐ fore, the decision was made to stratify the sample according to age, sex and social class. Another point of discussion is the sample size. Di Paolo & Yaeger-Dror (2010: 13) do not provide exact numbers for it but state that “there seems to be a general consensus in sociolinguistics today that relatively small judgement samples are sufficient.” They agree with Milroy and Gordon (2008: 29) in terms of practicality. Furthermore, depending on the outset and the goals of the study the number of recordings can vary. For example, Siebers (2007: 92) recorded 24 informants but chose only 16 speakers for a qualitative analysis of morphosyntactic variation in Black South African English while Baranowski (2008b: 531) analyses the 4.4 Sampling methods and contacting speakers 35 2 I am especially indebted to Ryan Whelpdale for his help. 3 I did not always analyse the speech of both persons because in some cases one person in the interviews did not speak enough or speech overlap occurred so often that not enough tokens could be extracted. speech of 43 speakers acoustically for his quantitative study of Charleston Eng‐ lish vowels. Since I was a stranger to Carlisle as community, and also because of time and budget constraints, I chose to visit venues where many people came together in order to approach potential interview partners. One way of meeting people was through attending Adult Education classes held in the evening where I intro‐ duced myself and asked for people’s assistance. Ten of the speakers in the sample were contacted in this way. They were either participants of classes or staff members. A second way of meeting people was through visiting community centres. The friend of a friend approach (Milroy 1987; Milroy and Gordon 2008) was successful in this context as well. Eleven people agreed to interviews after a friend of theirs had introduced me to them. 2 I also went to an Age Concern gathering in a residential home where I was also able to conduct some inter‐ views. During my time in Carlisle I stayed in St Martin’s College student halls (now University of Cumbria) and let everybody know on campus that I was doing research on Carlisle English. As a result, seven non-academic staff members offered me their help. I used several other methods, but they proved less successful: cold recruitment in shops, the library, approaching people at community meetings, posting ad‐ vertisements on Facebook and approaching the local fire station. 4.5 Final sample The sample of this study contains a total of 40 speakers, 20 females and 20 males. 19 speakers have a WC background and 21 have a MC background as defined above. Twenty-two of the recordings in the final sample are interviews with individual participants, in the other interviews speakers were recorded in dyads. 3 Where possible, the interviewees were recorded in their homes to make them feel as comfortable as possible. Sometimes the interviews took place at their workplace, in a (quiet) café or in a room provided by the Adult Education centre. Table 4.4 presents the distribution of speakers according to age, gender and social class. Unfortunately, not all cells could be filled equally. I aimed for four speakers per cell, but in particular young Working Class female speakers and 4 Methodology 36 middle-aged Middle Class male speakers are not as well represented. The im‐ balance probably stems from work commitments of the specific groups or even an oversight on my side about possible places where these groups could be contacted. Nevertheless, it should be possible to make claims on the basis of this data set. Male Female Total MC WC MC WC 22-39 2M26MC 3M31MC 4M34MC 1M22WC 5M36WC 6M38WC 2F23MC 3F24MC 4F25MC 5F25MC 7F36MC 1F22WC 6F27WC 13 40-59 7M41MC 11M57MC 8M43WC 9M52WC 10M55WC 12M59WC 8F42MC 9F47MC 11F53MC 10F49WC 12F56WC 13F57WC 14F59WC 13 60-78 13M60MC 14M61MC 17M67MC 19M73MC 20M78MC 15M62WC 16M63WC 18M68WC 15F60MC 18F67MC 19F68MC 16F61WC 17F63WC 20F71WC 14 Total 10 10 11 9 40 Table 4.4: Final sample used in this study 4.6 Interviews Sociolinguistic interviews are the most common method used to elicit speech data for language variation and change studies (e.g. Labov 1984; Baranowski 2007; Schleef 2013; Buchstaller 2015; Smith and Holmes-Elliott 2017). For the interviews, I used a digital recorder by Edirol with integrated stereo high-quality microphones. I did not use an external microphone because people could be too distracted by it and would probably pay more attention to their speech than desirable. The disadvantage is that the recording quality is not as good as when one uses an external microphone. The choice between very good recoding quality and the distraction caused by recorders or microphones is a balancing act, particularly in sociolinguistic interviews. The improvement of technical equipment in recent years has led to the overall improvement of recordings produced for sociolinguistic studies. The sound recordings today can have very 4.6 Interviews 37 4 5M36WC was recorded twice. During the first interview (March 2008) children were playing football in the adjacent room which made the recording unfeasible. For this reason I interviewed him a second time in September 2010. 5 Feagin (2004: 24) mentions that problems can occur when the interviewer is non-native to the culture and language. This was also the case for me. However, since I had spent some time in Britain before, I was not a complete stranger to the country. Me being new to the community automatically put me into “the position of the interviewer as a learner” (Labov 1984: 40). Thus, it was not awkward when I as a foreigner asked them about Carlisle. high quality even under fieldwork conditions. In the interviews for this study the recorder was placed unobtrusively in close proximity to the speaker(s), pro‐ viding satisfying data. The sampling rate at the time of recording was 44.1 kHz but it was down-sam‐ pled to 11 kHz for analysis. After the interview, the recordings were transferred to an Acer Aspire 5100 Notebook and also stored on data DVDs and on a hard drive. The acoustic analysis of the G O O S E , G OAT and F OOT vowel was done by using Praat (Boersma and Weenink 2008) and a complementary script (Kendall 2009) while the consonantal variation was analysed in ELAN. Not all recordings could be used for this study. Several of the recordings had to be excluded for the following reasons: • The recording quality for sociophonetic analysis was too poor due to background noise. 4 • Due to work commitments some interviews were too short to provide sufficient data. • I assume that one speaker ‘made up’ an accent with features no one else had. A native of Carlisle English cross-checked the recording and con‐ firmed my observation. 4.6.1 Interview procedure and oberserver’s paradox Before the interview, I told the participants that I was recording people for a project on the history and dialect of Carlisle, in the hopes that they would focus more on the history part rather than on language. Asking what it was like to grow up in Carlisle was often a good conversation starter. 5 People told me what they liked and disliked about living in Carlisle. The interview questions were prepared in advance in such a way as to elicit speech data. Yet the interview was only semi-structured. I considered the respective interview partner, and their age, when choosing interview questions. It was very important that the inter‐ 4 Methodology 38 viewees could relate to the topics and questions. Wolfram and Fasold (1974: 49) explain the aims of a sociolinguistic interview: The goal of spontaneous interviewing is quite straight-forward and simple: the inter‐ viewer wants to get as much free conversation as possible. He wants the informant to focus on the topic of his conversation so that he pays minimal attention to the way he is speaking. The less attention is paid to speech, the more informal and natural we can expect his speech to be (Wolfram and Fasold 1974: 49). A widely-recognised problem in social studies, e.g. sociolingusitics, psychology and sociology, is the observer’s paradox. Labov (1972b: 113) states that “to obtain the data most important for linguistic theory, we have to observe how people speak when they are not being observed.” Therefore, the overall goal of socio‐ linguists has to be to overcome the observer’s paradox when the speakers are recorded. “We must somehow become witness to the every-day speech which the informant will use as soon as the door is closed behind us: the style in which he argues with his wife, scolds his children, or passes the time of day with his friends” (Labov 1972b: 85). Hence, an important part of the sociolinguistic in‐ terview is to elicit casual speech. Labov (1984: 33) suggests employing the ‘danger of death’ questions, i.e. questions which potentially trigger emotional involvement and to elicit long narratives. While Tagliamonte (2006: 43) describes this type of question as working very well, many researchers have doubted its effectiveness. “The danger of death question does not translate well into every speech community” (Milroy and Gordon 2003: 66). For this study, the questions were adjusted to suit the context. Carlisle suf‐ fered heavy floods in 2005. At the time of the interviews, this was still fresh in people’s minds. Many interviewees shared their personal experiences, such as their house being flooded, often resulting in long stretches of narration. This topic proved to be a very successful ‘danger of death’ topic for the majority of speakers. Many older speakers also told me about a part of town called ‘The Lanes’ which had been demolished in the early eighties to be replaced by a shopping centre. 4.6.2 Style A reoccurring finding in sociolinguistic studies is intraspeaker variation in style shifting situations between casual and formal speech styles, where formal speech situations are defined as situations where speakers pay attention to their speech by self-monitoring. Labov (1972b: 208) argues that “styles can be ar‐ 4.6 Interviews 39 ranged along a single dimension, measured by the amount of attention paid to speech,” i.e. style-shifting from casual to formal speech situations by using re‐ sources such as word lists or reading passages. In the present study the inter‐ views usually included the following parts: • A reading passage containing 320 words, which was read by the par‐ ticipants of the study. The passage is about a young (non-British) person who - for the first time - spends a few months in Northern England. • A partly guided interview where participants were asked what it was like to grow up in Carlisle, what had changed in the city since their youth, how they liked the new shopping centre and whether they had been af‐ fected by the floods in 2005. Additionally, the interviewees were asked about their dialect, whether they had been mistaken for coming from a different dialect area, how they liked their dialect or accent and whether they would like to change anything about it. Other than that they told me about things they were interested in, their hobbies, travel, local history and their families. • A sentence list with 54 short sentences ranging from four to nine words. All lexical sets (except for PALM ) are represented in this list. The list is an adaptation from Hickey (2005). • The reading passage and the sentence list recordings were only analysed for the consonantal variables to detect style shifts. For the vowel variables the two reading tasks provided a first orientation. Not everybody com‐ pleted all these tasks, e.g. 6M38WC and 1F22WC decided not to do the two reading tasks. 4 Methodology 40 5 Sociolinguistic variables This chapter is concerned with the analyses of five sociolinguistic variables in the city dialect of Carlisle, G O O S E , G OAT and F O OT in chapter 5.1, (T) in chapter 5.2, (R) in chapter 5.3 and (TH) in chapter 5.4. The quantitative approach to language variation and change forms the analytical basis of this chapter. 5.1 GOOSE , GOAT and FOOT In this chapter the use of three variables in the community are reviewed: G O O S E , G OAT and F O OT . While G O O S E -fronting is a very prominent sound change in va‐ rieties of English, it is useful to include the analysis of G OAT and F O OT as well as parallel changes in relation to G O O S E are discussed for other varieties. Fronting describes a sound change in which the place of articulation of a vowel or a consonant moves from a back position towards a more front position in the oral cavity. Around the world, the fronting process of the high back vowel has received increasing interest in the last few years (e.g. Torgersen (1997) for Southern England; Hall-Lew (2009) for San Francisco; Mesthrie (2010) for South African English; Flynn (2012) for Nottingham English; Jansen (2018) for West Cumbria). As noted by Docherty (2010: 67) in an overview on recent changes in British English. Perhaps the most striking of these [vowel changes] is the fronting of the G O O S E and G O A T vowels by younger generations of speakers. Putting to one side those varieties where G O O S E is already fused with F O O T […], there are widespread reports of moderate to substantial fronting of G O O S E together with the production of much less marked lip-rounding/ protrusion […]. Ferragne and Pellegrino (2010), who analyse 13 accents across Britain, suggest a high variation in the G O O S E vowel across the country. But they also comment on the relationship between G O O S E and F OOT in the different varieties. These comments already hint at dynamics between the vowels G O O S E , G OAT and F O OT and they will be investigated in detail in this chapter. The first part of the chapter provides a short account of the development of the vowel represented by the lexical set G OO S E around the world and in England. I will also discuss in more detail the realisations of the G O O S E vowel in the tra‐ ditional Carlisle and Cumberland dialect. As a second step, I introduce the methodology for the acoustic analysis of G O O S E , G OAT and F O OT and present the results first of all for the three variables. In the discussion section I argue that the fronting of G O O S E in Carlisle English is the result of a diffusion process in England. However, the distribution of G O O S E -fronting in England suggests that the innovation does not stem from the Tyneside area. I also discuss the stability of G OAT and F OOT in relation to the changes observed in the G O O S E vowel. 5.1.1 GOOS E , G OAT and FOOT realisation in England Wells (1982: 147) defines the high back vowel in English as “the standard lexical set G O O S E [that is] comprising those words whose citation form in RP and GenAm has the stressed vowel / uː/ .” In the revised version of An Outline of English Phonetics, Jones in 1932 reports the fronting of / uː/ in palatal preceding environments in RP: The most important subsidiary long uː is an ‘advanced’ variety. It is used when j precedes, as in music ˈmjuːzik, tube ˈtjuːb, deluge ˈdeljuːdʒ. By calling it ‘advanced’ we mean that the part of the tongue which is highest is the central part - a part more forward than the ‘back’ ( Jones 1932: 82). Because of this quote we can assume that G OO S E -fronting (at least in the south of England) is a process which has been in progress for a century or longer. In 1982 Wells (1982: 148) mentions that a “rather central than back quality for G O O S E ” is found in “most English popular urban speech.” Thus, Wells recognises the fronting of the G O O S E vowel as common for urban vernaculars in England while Harrington, Palethorpe and Watson (2000) show that Queen Elizabeth II changes her pronunciation towards a fronter realisation of G O O S E between the 1950s and 1980s. Some general comments have been made about G OO S E -fronting in the north of England. Cruttenden (2008: 88) states that “many areas of Northern English have a fronted articulation of […] / uː/ .” Upton (2008: 272) agrees with Cruttenden in the synopsis of phonological variation in the British Isles that “the dominant realization of the G O O S E vowel is essentially [u(ː)] everywhere in the region, with fronting in varying degrees being a very common tendency.” Hence, a general fronting is observed in this geographical area but more information is not avail‐ able. 5 Sociolinguistic variables 42 1 Shorrocks (1998) also mentions other variants, which always include some kind of fronting. In one of the few comprehensive studies on varieties in the north-west of England, Shorrocks (1998: 170ff) describes the / uː/ vowel realised as [ʏː] 1 in Bolton, Lancashire. It is nevertheless “much less front, tense and rounded, how‐ ever, than Standard French or German / yː/ ” (Shorrocks 1998: 171). Beal (2008: 130) on the other hand does not discuss the G O O S E lexical set in her overview of Northern English phonology in much detail. Up until recently sociolinguistic studies of this change especially in the north of England were sparsely conducted but sociolinguists have now begun focus‐ sing on vowels with larger geographical spreading patterns (e.g. Flynn (2012); Hughes, Trudgill and Watt (2012); Jansen (2017a); Jansen (2018)). Beal, Burbano-Elizondo and Llamas (2012: 27) find a back G O O S E vowel in the speech of their representative male and female speaker from Newcastle who are both in their thirties. Beal (p.c.) argues that G O O S E -fronting could have been inhibited in Newcastle so far by high front rounded vowels found in the tradi‐ tional Northumberland dialect. However, Buchstaller (p.c.) and Warburton (2017) report G OO S E -fronting among adolescents in Newcastle in recent years. In addition to the fronting of G OO S E , Docherty reports on the fronting of G OAT in some English varieties in England, especially in the south. Varieties with monophthongal / oː/ for G OAT (e.g. Northern England) are not influenced by this innovatory fronting process. Only speakers in the south of England show the tendency for a fronter quality (Docherty 2010: 67). Historically F O OT and G O O S E are closely related vowel sounds (see Figure 5.1.1). Some words found in the F O OT lexical set belonged to the group of words with Middle English long / oː/ . After the Great Vowel Shift was completed (/ oː/ had risen to / uː/ ), a shortening in some of these words occurred and then joined the groups of words with Middle English / u/ , e.g. blood, good. In this group of words a phonemic split occurred between what are today the S T R UT and F O OT lexical sets with / ʌ/ and / ʊ/ qualities respectively. However, this split does not occur in the north of England. Thus, words such as sun and bus (belonging to the S T R UT lexical set) as well as foot and cook (belonging to the F O OT lexical set) are realised with / ʊ/ (Wells 1982: 198). 5.1 G O O S E , G O A T and F O O T 43 (i) (ii) (iii) Figure 5.1.1: Historical relationship between G O O S E and F O O T (taken from Wells (1982: 198)) Ferragne and Pellegrino (2010) address the relationship between the two lexical sets G O O S E and F O OT in present day data. They come to the conclusion that three main dispositions of G OO S E and F O OT are found in the British Isles: neither G O O S E nor F O OT seems to have moved from its back position […]; only G O O S E , but not F OOT , has moved to a fronter position […]; both G O O S E and F OOT are rather front […]. Type (i) is attested for East Yorkshire. The data was recorded in Hull. Here G O O S E and F O OT - S T R UT have a very back position compared to the other varieties. Fer‐ ragne and Pellegrino (2010) report that “there is nothing in our data that tells us that the change [fronting] will ever apply.” Newcastle English is also seen as comparably traditional in the realisation (i.e. G O O S E is fairly far back) but due to the small speaker sample, statements are only tentative for this variety. For the south of England, Ferragne and Pellegrino report the co-occurrence of fronting of G O O S E and F O OT which resembles type (iii). This result confirms Torgersen’s (1997) findings for the south-east of England. According to Ferragne and Pellegrino, type (ii) is found in several locations; one of them is Burnley in Lancashire which is (apart from Newcastle) the location closest to Carlisle fea‐ tured in this study (Ferragne and Pellegrino 2010: 18). These results from Lan‐ cashire can possibly give some implications for diffusional patterns and they 5 Sociolinguistic variables 44 2 If the vowel was not long enough for these measurements, measures were taken at 1/ 3 and 2/ 3 of the vowel. 3 This had mainly two reasons: On the one hand, you was often used in unstressed po‐ sitions and on the other hand the realisation of this word was in many cases / jɪ/ . 4 This was only the case for a few tokens. will be included in the discussion about the sound changes examined in this chapter. There has been a lot of discussion about the diffusion of consonantal features in England, Scotland and Wales in recent years. However, Kerswill (2003) while discussing the diffusion of consonantal feature, does not mention the diffusion of vowel variables across the country. Since vowel changes can happen gradually it is not easy to detect the diffusion of a change or a more or less parallel de‐ velopment across the whole country. 5.1.2 Data analysis Tokens of G O O S E , G OAT and F OOT were analysed in PRAAT (Boersma and Wee‐ nink 2008) with a complementary script based on (Kendall 2009). The script provides measurements at three points of the vowel, 0.035 seconds into the vowel, the midpoint and 0.035 seconds before the end of the vowel. 2 Tokens were only measured at primary stress occasions and the measurements were taken for the first, second and third formants at the midpoint. A total of at least ten tokens per speaker were collected. Only for one speaker fewer tokens, i.e. nine tokens, could be measured. Under certain circumstances tokens were omitted from the study, e.g. unclear production due to overlapping speech or background noise. You as token was dismissed from the sample. 3 Tokens which belong to the F OOT lexical set in the standard variety but were perceived as G OO S E in Carlisle English such as book or cook were not collected. 4 Because of possible coarticulatory effects, several restrictions were made concerning the phonological environments in which the vowel is measured, i.e. vowels following / w/ were omitted from this part of the study, / ŋ/ and / l/ when the vowel preceded (cf. Deterding 1997; Wells 1982). The previously mentioned statement by Jones (1932) hints at the possibility that the preceding allophonic environment has an influence on the degree of fronting of the high back vowel. In fact, Hall-Lew (2009: 160) comments that “phonological conditioning is particularly important for discussions of back vowel fronting.” Thus, the discussion of G O O S E -fronting must include internal factors such as preceding environment in order to asses this possible change. 5.1 G O O S E , G O A T and F O O T 45 In the linguistic literature a broad consensus exists about the environments which favour fronting of G O O S E and which do not. Nevertheless, in the details the studies vary. Mesthrie (2010) distinguishes between coronal, non-coronal and / j/ preceding environments while Baranowski (2007: 172) classifies between vowel-preceding coronals (Tuw), vowel-preceding non-coronals (Kuw) and / l/ following the vowel. Hall-Lew (2009) follows Flemming’s (2003) finding. He states that “not all types of coronals can condition vowel fronting [...] coronals that condition fronting are produced with a front tongue body, while coronals that condition retraction are produced with a back tongue body” (Flemming 2003: 336). According to Giegerich (1992: 117) the category of anterior coronal consonants is made up of / θ ð n t d s z l/ and is taken into account for the analysis of G O O S E here. Mesthrie (2010: 10) claims that “preceding consonant / j/ […] is a fronting environment par excellence and deserves to be isolated as a special fronting environment”, tokens with this particular environment will be grouped together. In the present study, tokens with preceding palatal are included in order to detect fronting tendencies. As yod-dropping does so far not occur in Carlisle English, the danger of varying preceding conditions in this subset is not given. The following preceding environments were included in the analysis: vowels following anterior coronal consonants ( TO O ), vowels following palatal / j/ ( N EW ) and vowels following any other environment ( C O O P ). Table 5.1.1 summarises the details of the split-up of G OO S E into the three subsets with the linguistic infor‐ mation and example words. Lexical set Details Examples G O O S E preceding anterior coronal ( T O O ) lose, two, soon preceding palatal / j/ ( N E W ) news, Tuesday preceding other ( C O O P ) goose, boot, shoe Table 5.1.1: Distribution of lexical subsets of G O O S E in this study Altogether, the number of tokens analysed for this chapter was 2,858, including 1,242 tokens analysed for G O O S E (males: 565; females: 677). The G OAT and F O OT data sets are not divided any further into different phonological environments with the G OAT data set containing 1,025 tokens and the F O OT data set containing 591 tokens. 5 Sociolinguistic variables 46 5.1.3 Normalisation Normalisation methods are algorithms which are applied to the phonetic data in order to make it comparable across speakers and speaker groups. Linguistic studies exist which discuss and present vowel plots of non-normalised vowels (cf. Watt and Tillotson 2001), but in order to compare single speakers or speaker groups normalisation is crucial. Physiological differences such as the length or height of the oral cavity which influence the formant frequencies are limited through the normalisation procedure. At the same time normalisation methods should retain accent differences (Thomas 2004: 168). “Thus for […] studies in which vowel quality of different speakers is compared, normalization is crucial” (Thomas 2007: 221). The overall goals of vowel normalisation methods should be: a. to eliminate variation caused by physiological differences among speakers; b. to preserve sociolinguistic/ dialectal/ cross-linguistic differences in vowel quality; c. to preserve phonological distinctions among vowels; d. to model the cognitive processes that allow human listeners to normalize vowels uttered by different speakers (Disner 1980 quoted in Thomas and Kendall 2007; Flynn 2011). Watt, Fabricius and Kendall (2010: 113-115) categorise normalization methods into two main groups: vowel-extrinsic and intrinsic; and speaker-extrinsic and intrinsic methods. Vowel-intrinsic methods normalise data according to the in‐ ternal structures of the vowel so that only the lexical set in question is needed for the normalisation procedure, e.g. in this study only tokens for the lexical sets G OO S E , G OAT and F O OT (the ones that are analysed here) would have to be nor‐ malised. When vowel-extrinsic methods are applied, tokens from the lexical set in question as well as (at least) lexical sets representing peripheral vowels in a phonological system have to be collected in order to represent the data in rela‐ tion to the other vowels. Speaker-extrinsic methods take the complete group of speakers into account while in speaker-intrinsic methods data is normalised within a speaker’s system. Fabricius, Watt and Johnson (2009) put forward a modified version (mW&F) of their vowel-extrinsic and speaker-intrinsic normalisation method which is based on the earlier version W&F (Watt and Fabricius 2003). This normalisation method has been applied in various studies (Kamata 2006; Fabricius 2007; Mes‐ thrie 2010). They summarise the mechanisms of their method as the following: “The S-procedure normalizes a speaker’s set of vowel data by expressing each 5.1 G O O S E , G O A T and F O O T 47 5 The W-F ratios are based on the values extracted from the W&F method (or mW&F). formant value as a proportion of its respective centroid value, which is derived using F1 and F2 maxima and minima for that individual’s vowel space” (Fabri‐ cius, Watt and Johnson 2009: 414). In this study, the acoustic measurements in Hertz are normalised with the modified Watt & Fabricius method (Fabricius, Watt and Johnson 2009) in order to make the data comparable across the speakers. Due to the vowel-extrinsic nature of this method, in addition to the lexical sets analysed here, token meas‐ urements for the following lexical sets were collected and normalised in addition to the variables analysed in this chapter: F L E E C E , THO U GHT , LOT , BATH , S TAR T , S T R UT and P R IC E . These vowels were chosen in order to provide tokens for pe‐ ripheral vowels. 5.1.4 Classification of fronting Similar to Baranowski (2007: 171-203), in this study the normalised F2 values of each speaker are plotted in order to detect language change. Mesthrie (2010) points out that the approximate position in the vowel space traditionally rep‐ resented by a vowel chart is not defined by the normalised F2 value and he tries to fill this gap. He uses the mW&F method, similar to the present study, as means of normalisation. In his opinion a general scale with traditional values should be implemented for the vowel chart: […] since sociolinguistics also draws extensively on traditional phonetic terminology, it would be useful to match the W-F ratios, 5 however roughly, with traditional posi‐ tions in a vowel chart. No such mapping currently exists and it is a desideratum for the future that such a mathematical transform be developed (Mesthrie 2010: 11). He suggests a scale for high vowels in order to classify their degree of fronting (see Figure 5.1.2). However, this scale cannot be applied to more open vowels due to the physical properties of the oral cavity. Algorithms which could deter‐ mine the position of more open vowels have not been developed. According to Mesthrie (2010), the scale captures the approximate position of the high back vowel and is applied as an indicator for the relative position in this study. The scale operates between the extremes 0 and 2.0 (the mW&F normalisation scale operates between 0 and 2.0) where the value 1.0 denotes a central position and the other traditional values are organised around this value. The normalised F2 values for G O O S E extracted in this study are assigned to these more descriptive values in a second step. 5 Sociolinguistic variables 48 Figure 5.1.2: Scale for fronting proposed by Mesthrie (2010: 12) for data normalised with mW&F 5.1.5 Statistical modelling Statistical modelling has become ever-increasingly important in sociolinguistic and sociophonetic studies. In the present study, I use Rbrul software ( Johnson 2009) as a statistical toolkit which runs logistic regressions taking mixed effects into account in order to identify predictors of language change. Mixed effects models can take into account by-speaker and by-item correlations. They estimate between-group effects (like sex) at the same time as within-group effects (like individual speaker). They support continuous factor groups like age or lexical item as well as continuous responses (dependent variables) like vowel formant measurements. This means you can include all kinds of different factors in the same model (Taglia‐ monte 2012: 138). Thus, the advantage of the application of this statistical tool is to include random factors such as individuals and also age as a continuous variable. With more traditional tools age has to be categorised and the individual is disregarded. The analyses were based on normalised F2 values as dependent variable. The social factors of age, sex and social class and the linguistic factor environment which were entered into the statistical model were treated as independent var‐ iables and each speaker as random factor. The speaker’s age was treated as a continuous variable based on the age of the speakers when the interview took place and the individual speaker was included as random factor. Male and female constitute the binary distinction in the social factor sex. Social class is also de‐ fined as a binary choice MC vs. WC on the basis of educational and occupational information. 5.1 G O O S E , G O A T and F O O T 49 5.1.6 Results 5.1.6.1 Overall profile of the data for G O O S E Table 5.1.2 presents the normalised F2 values for each speaker according to age. The average values for each speaker are then categorised according to the scale by Mesthrie (2010: 12). The normalised mean F2 values range from 0.893 to 1.578 where higher F2 values indicate greater fronting. In general, the results suggest that fronting of the high back vowel is under way in the community. Speaker Age Sex Social Class F2 G O O S E Level of Front‐ ness 1F22WC 22 F WC 1.317 frontish 1M22WC 22 M WC 1.4 front 2F23MC 23 F MC 1.472 front 3F24MC 24 F MC 1.567 front 4F25MC 25 F MC 1.578 front 5F25MC 25 F MC 1.537 front 2M26MC 26 M MC 1.564 front 6F27WC 27 F WC 1.369 frontish 3M31MC 31 M MC 1.437 front 4M34MC 34 M MC 1.262 frontish 7F36MC 36 F MC 1.387 frontish 5M36WC 36 M WC 1.355 frontish 6M38WC 38 M WC 1.229 frontish 7M41MC 41 M MC 1.332 frontish 8F42MC 42 F MC 1.451 front 8M43WC 43 M WC 1.239 frontish 9F47MC 47 F MC 1.39 frontish 10F49WC 49 F WC 1.213 frontish 9M52WC 52 M WC 1.066 central 11F53MC 53 F MC 1.26 frontish 10M55WC 55 M WC 1.03 central 5 Sociolinguistic variables 50 12F56WC 56 F WC 1.485 front 13F57WC 57 F WC 1.32 frontish 11M57MC 57 M MC 1.156 central 14F59WC 59 F WC 1.133 central 12M59WC 59 M WC 1.176 central 15F60MC 60 F MC 1.316 frontish 13M60MC 60 M MC 0.893 backish 16F61WC 61 F WC 1.155 central 14M61MC 61 M MC 1.061 central 15M62WC 62 M WC 1.051 central 17F63WC 63 F WC 1.077 central 16M63WC 63 M WC 0.897 backish 17M67MC 67 M MC 1.318 frontish 18F67MC 67 F MC 1.184 central 19F68MC 68 F MC 1.09 central 18M68WC 68 M WC 0.96 central 20F71WC 71 F WC 0.943 central 19M73MC 73 M MC 1.119 central 20M78MC 78 M MC 1.233 frontish Table 5.1.2: Average F2 values of G O O S E represented by the normalised average mW&F score for each speaker and its approximate position in the vowel chart according to Mesthrie (2010) As a second step I ran a logistic regression. Table 5.1.3 provides the statistical results for the lexical set G OO S E . The first four lines of the table provide us with descriptive statistical information. The deviance tells us how much the actual data deviates from the predictions of the model. The fairly low negative number of the deviance gives us the information that the model fits the data quite well. The statistical results for the external and internal factors are discussed below. The degrees of freedom (df) provide us with the information how many factors 5.1 G O O S E , G O A T and F O O T 51 are included in the model and the grand mean provides the overall mean of F2 values in this model. Deviance -628.591 df: 8 Grand mean: 1.274 Speaker ID random standard deviation: 0.099 Applications Value: F2 Value of G O O S E Factors Log Odds Tokens (N) Mean Value Age p<0.01 Continuous 22-78 +1 -0.008 Sex p<0.05 F 0.044 677 1.327 M -0.044 565 1.211 Social class p<0.05 MC 0.049 660 1.342 WC -0.049 582 1.198 Environment p<0.01 N E W 0.107 425 1.357 T O O 0.001 571 1.274 C O O P -0.108 246 1.134 Table 5.1.3: Rbrul output results for G O O S E based on the F2 measurements of all G O O S E tokens 5.1.6.2 Social factors According to the apparent time hypothesis, language change can be observed when you study different age groups at a given point in time. Thus, the way that younger speakers realise the G O O S E vowel should consistently be different from older speakers. Indeed, the results in the boxplot in Figure 5.1.3 suggest a correlation between the fronting of G O O S E and age. The figure provides the nor‐ malised average F2 values for all the speakers. The x-axis represents the speaker age from young to old while the y-axis represents the normalised average F2 values. There is a clear trend of increasing F2 values for younger speakers. This observation is supported by the statistical results (p<0.01; see Table 5.1.3). This 5 Sociolinguistic variables 52 result suggests a change in progress. The regression lines indicate the trajectory of the change. Figure 5.1.3: Fronting in the G O O S E vowel according to the speakers’ age and sex What is also immediately clear from Figure 5.1.3 above is that both sex groups take part in the change and that there is a sex difference in the distribution of normalised average F2 values. The data shows that females generally realise F2 higher (average: 1.327, i.e. their articulation of G O O S E is more fronted) than male speakers (average: 1.211) but both groups participate in the change towards fronting. This result is significant at the p<0.01 level. The results for social class show that both groups - WC and MC - take part in the change towards more fronted G O O S E values where the group of MC speakers have a higher average F2 value (1.342) than WC speakers (1.198). Hence, as well as age and sex, social class is a predictor for the extent of fronting at the p<0.05 level. In particular female MC speakers tend to produce the G OO S E vowel in a more front position (i.e. the F2 value is higher) than WC speakers, but for the young male speakers the distinction between WC and MC seems to collapse. The fact that young women are leading this change (Figure 5.1.3) is in line with other findings concerning supralocalisation. Women have been identified as very important in promoting supralocal changes. Flynn’s (2012) findings sug‐ 5.1 G O O S E , G O A T and F O O T 53 gest that sex differences hardly occur for G O O S E fronting in his sample of Not‐ tingham English. Even though this change is most likely a supraregional change, females do not take a leading role in it in Nottingham but they do in Carlisle. The results of the logistic regression presented in Table 5.1.3 confirm that the three external factors age, sex and social class as well as the internal factor environment are significant factors for this change. The highly significant results across all factors lead to the conclusion that this change is still ongoing and that all groups are involved. Figure 5.1.4 and Figure 5.1.5 present vowel spaces of two speakers of Carlisle English at the extreme ends of the vowel continuum. The vowel plot of a 60-year-old MC male speaker is presented in Figure 5.1.4. G O O S E still occupies a back position for this speaker even though the standard deviation is quite high and therefore variation within the speaker is quite high. The speaker still has a more traditional realisation of the G OO S E vowel even though the average of F2 values for G OO S E is produced more fronted than the F O OT vowel. Yet the standard deviation reveals that there are also instances where G O O S E is produced further back than F O OT . The high deviation and variation within the speaker could be interpreted as a first step towards a change, though still without an obvious fronting. The speaker chooses from a wide range of realisations but the indi‐ vidual threshold for constant fronting has not been passed. Figure 5.1.4: Vowel plot of a 60-year-old MC male speaker (13M60MC) 5 Sociolinguistic variables 54 Figure 5.1.5 represents the vowel space of a 23-year-old female MC speaker. It is apparent from this figure that speaker 2F23MC participates in the vowel change where G O O S E is found in a very front position. The deviation is much lower for her than for the speaker represented by the vowel space in Figure 5.1.4. This might hint at a deceleration of the change. There is also already some overlap with the vowel space of the F L E E C E vowel. What is also apparent from this chart is that F O OT and G OAT are found in a back position for this young person. Thus, she is taking part in the fronting of G O O S E but for the other two vowel sets a back quality is still found Figure 5.1.5: Vowel plot of a 23-year-old MC female speaker (2F23MC) 5.1.6.3 Linguistic factors Turning now to the linguistic factors which possibly influence the fronting of the G O O S E vowel, Table 5.1.4 presents all speakers and their normalised F2 values for the G O O S E subsets TO O , C O O P and N EW . A difference in distribution between the three lexical subsets N EW , TO O and C O O P can be observed in Figure 5.1.6 below. Environment is highly significant as a predictor for fronting (p<0.01) (see Table 5.1.3). These findings are similar to the ones found by others, e.g. Baranowski (2007); Hall-Lew (2009); Mesthrie (2010). However, even though differences in the average values are found for 5.1 G O O S E , G O A T and F O O T 55 the three subcategories, all subcategories are involved in the change towards higher F2 values and therefore to more fronted variants. Figure 5.1.6: Fronting of the G O O S E vowel according to environment Table 5.1.4 illustrates the distribution of the average F2 values across environ‐ ments according to the positions they occupy in the individual vowel space according to the scale suggested by Mesthrie (2010). From the overall distribu‐ tion it is clear that the N EW environment is indeed the environment where one finds most overall average values in the ‘front’ space and least of the ‘front’ values in the C O O P environment. On the other hand, in the C O O P environment more ‘central’ and ‘backish’ vowel values are found than in the other two sub‐ sets. These results emphasise that internal constraints are at work in this change. N EW T O O C O O P front 19 8 5 frontish 12 19 11 central 11 14 22 backish 0 1 3 5 Sociolinguistic variables 56 back 0 0 0 Table 5.1.4: Distribution of average F2 values across approximate positions according to Mesthrie (2010) Only four speakers have a ‘backish’ vowel quality and none of the speakers has a G O O S E vowel that is categorised as ‘back’, therefore G O O S E might never have been a truly back vowel or fronting had already begun before the oldest speakers of this sample were on the ‘linguistic market’ and because of that none of the speakers occupies this very back area in the oral cavity. A problem that occurs with this classification is the fairly high number of ‘front’ qualities. The auditory judgement of Carlisle English and other (more Southern English) varieties leaves one with the impression that the area that is defined as ‘front’ must represent a wide range of vowel qualities because Carlisle English speakers with ‘front’ qualities produce values that are by far not as front as e.g. Bolton English speakers. I argue that even though a scale which combines acoustic measure‐ ments with the traditional vowel space approach is desirable, Mesthrie’s scale still needs some refinement. 5.1.6.4 Results for G O A T In this part of the chapter I focus on possible variation in the G OAT lexical set. Deviance: -2026.286 df: 5 Grand mean: 0.981 Speaker ID random standard deviation: 0.046 Applications Value: F1 Value of G OAT Factors Log Odds Tokens (N) Mean Value Age p<0.05 Continuous 22-78 +1 -0.001 Sex p<0.05 M 0.041 493 1.022 F -0.041 532 0.942 Table 5.1.5: Rbrul output results for G O AT based on the F1 measurements of all G O A T tokens 5.1 G O O S E , G O A T and F O O T 57 Deviance: -2134.887 df: 5 Grand mean: 0.758 Speaker ID random standard deviation: 0.081 Applications Value: F2 Value of G O AT Factors Log Odds Tokens (N) Mean Value Sex p<0.01 M 0.044 493 0.797 F -0.044 532 0.722 Social Class p<0.01 MC 0.036 510 0.787 WC -0.036 515 0.730 Table 5.1.6: Rbrul output results for G O A T based on the F2 measurements of all G O A T tokens The statistical results for F1 (Table 5.1.5) suggest that male speakers have a higher average F1 value (1.022) than female speakers (0.942). Higher F1 values propose a lower position in the vowel space, i.e. male speakers use a lower (more open) vowel than female speakers. In addition, age is a significant factor. The change moves from a higher to a lower F1 value, i.e. a higher (or closer) vowel [oː] which displays a northern supralocal form. For the dependent factor F2 the statistical results provide us with very im‐ portant information as well (Table 5.1.6). Even though the social factors sex and social class are significant for F2, age is not. Hence, a change in progress, i.e. fronting is not observable in Carlisle English. However, the difference in the vowel value distribution lets us assume that higher F2 values (= fronter realisa‐ tion of the vowel) are indexical for male speakers and lower F2 values are in‐ dexical for female speakers. The same is true for MC and WC speakers where the former group has higher F2 values. There are several reports about G OAT fronting in the north of England now, in particular in Yorkshire (e.g. Watt and Tillotson 2001; Finnegan 2009; Hughes, Trudgill and Watt 2012; Cooper 2017). However, the results for Carlisle English attest that this change is not underway in the community. I argue elsewhere ( Jansen 2017a) that the phonological setup of vowels in Carlisle and other northern communities prevents the fronting of G OAT . G OAT needs to have a diphthongal quality in order to participate in the fronting process which ties in with Docherty’s (2010) statement. 5 Sociolinguistic variables 58 Besides the structural reason, attitudes could play a role as well. The very back quality of G OAT distinguishes Carlisle English speakers from people from Newcastle where a fronted quality is one of the traditional variants. Jansen (2013) shows that there are quite striking differences on the segmental level between Carlisle and Newcastle English and it might be a conscious choice for Carlisle people to differentiate themselves from the north-east. However, it seems that G OAT as a back vowel has reached a higher indexical order (cf. John‐ stone and Kiesling 2008: 8f). In various forums in social networks people from Carlisle have mentioned that it was hard to order a coke outside of Carlisle. This is probably due to the very back quality of G OAT in the community while the G OAT vowel is shifting more and more towards a more fronted quality in other varieties which possibly makes it a salient feature to other groups. These groups comment on the use of the G OAT vowel which makes the feature salient to Carlisle English speakers. 5.1.6.5 Results for F O O T Statistical analysis shows that neither age, nor sex or social class predict possible fronting. Therefore, for the lexical set F O OT no significant fronting is observed which could be interpreted as change in progress. 5.1.6.6 Summary G O O S E - G O A T - F O O T The analysis of the vowels has shown that G O O S E is fronted in all three lexical subsets and that G OAT and F O OT have comparatively stable F2 values. The results presented in this chapter indicate that with the fronting of G OO S E a sound change is underway that carries social meaning. Another important finding is that neither the G OAT vowel nor the F O OT vowel are fronting in Carlisle. The change towards a fronted quality of F OOT does not seem to affect the north and G OAT fronting only occurs in varieties where the vowel has a diphthongal quality (cf. Haddican et al. 2013) which is not observable in Carlisle English. The lack of the F O OT - S T R UT split seems to prevent the diffusion of F OOT -fronting in the north (cf. Jansen 2017a). 5.1.7 Discussion This chapter shows that G OO S E -fronting - as in many other varieties - is a change in progress in Carlisle. Younger speakers produce more fronted nuclei than older speakers and young MC female speakers are taking the lead in this change. Logistic regressions show age is statistically significant for the devel‐ opment of F2 in G O O S E and in all three lexical subsets of G OO S E in apparent-time 5.1 G O O S E , G O A T and F O O T 59 6 This finding is under reserve. Current sociolinguistic studies have not focused on this variable in Newcastle English which could give more details. while F2 of G OAT and F OOT is fairly stable. The different allophonic environments are considerably influencing the degree of fronting, a result which is comparable to other studies, e.g. Baranowski (2008) and Mesthrie (2010). Similar to their findings, N EW tokens have higher F2 values than TO O and C O O P . Anterior coro‐ nals preceding the vowel are influencing the degree of fronting as well (as pre‐ dicted by Flemming 2003). For the F2 values of G OAT a social stratification of sex and social class is ob‐ servable but there is no obvious change in progress because age is not a signif‐ icant factor. The resistance to fronting of G OAT possibly stems from system-in‐ ternal conditions, i.e. diphthongisation needs to precede the fronting process and the fact that the very back G OAT vowel is a feature in Carlisle English that is distinctive from the fronted quality of the vowel in traditional Newcastle English. According to Ferragne and Pellegrino’s (2010) model mentioned above, Carlisle English resembles type (ii) in the tripartite system with fronting for G OO S E but not for F O OT . As discussed earlier, Newcastle English is still said to have a more back quality of the G OO S E vowel. 6 Based on this assumption, a dif‐ fusional change from the north-east to Carlisle is therefore less likely. A similar distribution for G OO S E and F OOT as in Carlisle English is found in Burnley, Lancashire (Ferragne and Pellegrino 2010). Hence, diffusional change from the north-western regions of Lancashire, Greater Manchester (cf. Shor‐ rocks 1998 for Bolton), Merseyside or Cheshire rather seem to be a trigger for the high back vowel fronting process in Carlisle. This direction of diffusional change is slightly surprising since the very rural Lake District and the rural Lancashire area are not the predominant areas people from Carlisle come into contact with, i.e. there is no urban area and the Lake District almost functions as a barrier between the northern part of Cumbria and Lancashire. Nevertheless, the occurrence of a similar pattern in Burnley is most likely not a coincidence. Close proximity to Scotland could lead to the assumption that the fronting of the high back vowel is a result of dialect contact with this variety. There are however reasons why this hypothesis has to be rejected. First of all, Watt, Llamas and Johnson (2010) show that contact features of Scottish English are hardly found in Carlisle English. In addition, dialect contact of Carlisle English with Scottish English could not explain this change, since the former shows phono‐ logical length distinction for vowels (including G OO S E and F OOT ) whereas the latter does not. Words that resemble the F O OT lexical set in Carlisle would have 5 Sociolinguistic variables 60 to show fronting in comparable measures and be merged with G O O S E . Since this is not the case but rather that F O OT stays in a back position in Carlisle, the hypothesis of an influence from Scottish English must be discounted. Phoneti‐ cally, another argument has to be taken into account. A difference between the rounded Scottish English vowel and the more unrounded Carlisle G O O S E vowel exists. Hence, there are phonetic and phonological reasons why Scottish English cannot be the source of fronting observed in Carlisle English. Across English varieties, the high back vowel seems to be fairly unstable. Torgersen and Kerswill (2004: 45) comment on the fronting of this vowel in the context of chain shifts. The existence of these parallels may suggest an internally motivated explanation of the vowel changes, that is, a chain shift according to the principles described above. Clearly, the fronting of F O O T (and that of G O O S E and G O A T ) is widely observed in present-day English throughout the world. Given the very great differences in the sociolinguistic set-ups, as well as in the vowel systems, of the varieties concerned, we can safely say that there is something ‘natural’ about these changes, neatly encapsu‐ lated in Principle III, even though the presence of a chain shift seems not to be a precondition. This is possibly the case for Carlisle English as well. There is an internal dispo‐ sition for the instability of the high back vowel (which stretches across varieties of English). Henton (1983) sees the reason in the lack of / y/ in (RP) English which permits the fronting of / u/ and in fact, in German where / y/ exists along / u/ , high back vowel fronting is not observed. Certainly sociolinguistic factors and possibly diffusional patterns as discussed above also play a role in this change. 5.1.8 Conclusion This chapter analysed the production of the high back vowels ( G OO S E ) and in relation to the mid back vowel ( G OAT ) and the near close near back vowel ( F O OT ) in Carlisle English. Diffusional forces are at work in this variety for the high back vowel but for the other two vowels fronting processes are not observed. Dialect contact with Scottish English - which is a quite frequent contact scenario - does not trigger this change but - most likely - the contact with people from the Lancashire region. Nevertheless, diffusion cannot be the only reason for a change which occurs in so many varieties of English around the world; an in‐ ternal disposition or instability of the G O O S E vowel is a precondition for this change. 5.1 G O O S E , G O A T and F O O T 61 7 Liverpool English is an exception. 5.2 (T) The second variable to be discussed here is the variation in (T). This variable has been analysed for more places in recent years, e.g. (Reid 1978; Mees 1987; Trudgill 1988; Milroy et al. 1994; Stuart-Smith 1999b; Sangster 2001; Watson 2006; Jones and Llamas 2008; Drummond 2011; Schleef 2013; Smith and Holmes-Elliott 2017; Jansen submitted). This is mainly due to the diffusion of the glottal stop as one variant of (T) which is associated with London and the central belt in Scotland. The use of glottal stops has been covered extensively in the media. Foulkes and Docherty (1999a: 11) claim that “glottal(ised) forms of voiceless stops […] have become so ubiquitous as to generate regular (and almost always unfav‐ ourable) comments in the media.” Milroy et al. (1994: 350) mention that “it is clear from a number of independent reports that glottalisation phenomena are currently involved in rapid linguistic change in locations in different parts of Britain, and that British English is rather generally affected by this change.” A discussion of (T) in the Carlisle English context can add to the information about the distribution of glottal stops of non-initial (T) because of the distance of Carlisle to London. Moreover, the presence of [ʔ] in the Carlisle data will lend further support to the observation that the variant is becoming an accepted feature in dialect areas some distance away from the capital. This is not to say that London is necessarily the place where this change originated (see Schleef 2013 and Smith and Holmes-Elliott 2017 for further discussion). This chapter discusses the distribution of glottal stops in Carlisle English. Even though the glottal stop does not belong to the English phonemic inventory, phonetically, this allophone of (T) is found in many English 7 dialects in the UK. Unfortunately, terminology is very fuzzy. Glottaling and glottalisation, glottal reinforcement, glottal stops and glottal replacement are all used more or less synonymously even though some of these terms do not describe the exact same articulatory process. In this chapter the term used for the use of glottal stops is T-glottaling, glottal stop or glottal replacement invariably. 5.2.1 Historical and geographical background When it comes to identifying the origin of glottal stops in English dialects, the traditional Cockney accent is often mentioned as the source: “Preglottalization and glottaling are found in many kinds of English. Yet the glottal stop is widely 5 Sociolinguistic variables 62 8 Phonetically this seems to be a strenghthening with one oral and one glottal stop in a row while the sole use of glottal stops is rather a phonetic weakening due to the removal of the oral gesture. Glottalisation, i.e. the combined use of oral and glottal stop is still found in the north-east, e.g. in Newcastle-upon-Tyne and Sunderland. regarded as a sound particularly characteristic of Cockney. […] A bare [ʔ] as the realization of word-internal intervocalic / t/ is one of the most stereotyped char‐ acteristics of Cockney […]” (Wells 1982: 323f). A second urban area where the use of glottal stops is already attested in the middle of the 19 th century is Glasgow. Stuart-Smith (1999b) and more recently Schleef (2013) and Smith and Holmes-El‐ liott (2017) mention the historical component of this feature in Scotland. In par‐ ticular, WC speakers at the Scots end of the continuum between Scots and Standard Scottish English use this variant. However, the use of glottal stops was not restricted to these two areas in the 19 th century. Beal (2007) shows that the feature is described for the north of England in the 19 th century on the basis of literary representations of dialect. She quotes Jespersen (1909) who makes this observation during his travels in Britain in 1887: The glottal stop [’] produced by a closure of the vocal cords […] is not a regular element of the English sound system […]. After a vowel it is found pretty often in the North of England and in Scotland, especially among the uneducated, but by no means ex‐ clusively among dialectal speakers […]. I have heard it in the following words: in Sheffield tha’t (very often), can’t, thin’k, po’pe, boo’k; in Lincoln I’ts, migh’t, cer’tainly, u’p, wha’t, bough’t, thin’k, si’t; in Glasgow don’t, wan’t, o’pen, go’t, tha’t, brigh’tening, no’t; in Edinburgh in a great many similar words. Sometimes we have the further development that the mouth stop is omitted, as in [wɔ’er] for water (Edinburgh) ( Jes‐ persen 1909: 414; quoted in Beal 2007: 40). She speculates that the use of glottal stops diffused from the Glasgow area to the north of England in the 19 th and early 20 th century (Beal 2007: 40) 8 and sug‐ gests that a “pincer movement” (Beal 2010: 79) was involved in the diffusion of glottalisation. Thus she argues against Kerswill’s hypothesis (Kerswill 2003: 232) that glottaling is an innovation which has diffused from the south-east in the 20 th century. Andrésen (1968: 18) presents evidence that glottal replacement of intervo‐ calic / t/ first appeared in the west of Scotland (including Glasgow) around 1860, spreading to the east of Scotland and the far north of England some years later. According to Andrésen, the first evidence of glottal replacement in the London area is from around the beginning of the 20th century, glottal replacement and glottal reinforcement of intervocalic / t/ was a feature of rural dialects in most 5.2 (T) 63 eastern England, but not the southwest, the Midlands or northern England, in‐ cluding Yorkshire (Trudgill 1974: 81). The feature seems to have diffused to urban centres outside the southeast within the last 30-40 years (see Mees 1987 on Cardiff and Watt and Milroy 1999 on Tyneside). The phenomenon is therefore considerably older in the southern towns than it is in Hull, which lies in the traditional East Riding of Yorkshire (Kerswill 2003: 232). As for Carlisle English, the use of glottal stops has not been mentioned spe‐ cifically. However, Wright (1979: 33) reports T-glottaling for Cumbria: Lastly, what shall we say about the glottal stop, that sounds like a little cough mas‐ querading as a t in words like pi’y (pity) and Wukki’n (Workingtown)? Some would reply, ‘The less said, the better.’ They brand the glottal stop as ‘harsh, lazy, ungainly, horrible’; but these are risky value judgments because the poor glottal stop is only a sound, and one that is standard feature of some languages such as German. However, keeping to factual reporting, it is noted here that the glottal stop has reached Cumbria, most strikingly in the speech of youngsters in Barrow, where it has probably spread from Lancashire towns. To a smaller extent it is reaching the younger generation of Whitehaven and Workington, but has not spread to rural districts. In fact, compared with most English shires (outstanding among which in this respect is County Durham) and cities (for example Brum, Leicester and above all London), Cumbria is still fairly free of the glottal stop (emphasis: mine). Wright seems to describe a very early stage of the use of intervocalic T-glottaling in the county. He compares the use of glottal stops in Cumbria - without any quantitative data - with the use of this variant in larger cities and comes to the conclusion that it is not as common in Cumbria as it is in other cities. He seems to describe a diffusion pattern of T-glottaling from the south (Barrow-on-Fur‐ ness) up north which makes the possibility that Cumbrian use of T-glottaling was influenced by Scotland or the north-east rather unlikely. Furthermore, Wright (1979) mentions what can be interpreted as a cascade model for the use of the variant [ʔ] for / t/ where it reaches the towns first before it is used in rural areas. 5.2.2 (T) in Britain As indicated above, T-glottaling has been one of the most documented features of English varieties in the British Isles in recent years. It is a feature which is found in the vernacular of the majority of urban varieties nowadays (Milroy et al. 1994; Torgersen and Kerswill 2004; Britain 2005). The collection of chapters 5 Sociolinguistic variables 64 on urban areas in the British Isles (Foulkes and Docherty 1999b) reveals the spread of this feature: • Derby: “[ʔ] is almost categorical for word-final / t/ before a consonant. It is also the majority pronunciation where / t/ occurs pre-pausally, but sig‐ nificantly more so for young speakers […] The most marked sociolin‐ guistic patterning occurs in pre-vocalic position […] where glottals occur only 9% of the time for older speakers, in contrast with 60% for younger speakers” (Docherty and Foulkes 1999: 50). • Sheffield: “[ʔ] is used for non-initial / t/ much more frequently by younger speakers” (Stoddard, Upton and Widdowson 1999: 75). • West Wirral: “Pre-consonantal/ final / t/ was often glottalised as it is in‐ creasingly in many areas” (Newbrook 1999: 97). • Sandwell: “The glottal stop is very frequent in teenage speech and also variably in young adult (30 yrs) speech … but very infrequent in the speech of the elderly” (Mathisen 1999: 110). • Hull: “Older speakers in Hull use less glottal replacement of / t/ than younger people” (Williams and Kerswill 1999: 147). • Glasgow: “Glottaling of non-initial / t/ is a strongly stigmatized yet ex‐ tremely common feature of Glaswegian. […] An initial investigation of the 1997 data reveals a significant increase in glottaling, and continued social differences in patterning according to phonetic environment […]” (Stuart-Smith 1999a: 208f). Kerswill (2003) explains that the spread of the glottal stop as a variant of / t/ has been part of the geolinguistic process of diffusion which originated in London. He describes consonants as “torchbearers of change”, i.e. mainly consonantal variables are spreading across the country. Articulatory details about the glottal stop provide additional information about these variants. By definition a glottal stop is produced as a stop in the glottis. If [ʔ] replaces [t] for / t/ we talk about a retraction or backing of the place of articulation. However, with the advance of acoustic linguistic tools, recent research has shown that creaky voice is another way to produce a perceived stop (Docherty and Foulkes 1999; Docherty and Foulkes 2005). We hesitate to say glottal stop, since acoustic analysis shows that sounds perceived as [ʔ] or an oral plosive with glottal reinforcement in fact tend not to contain a glottal stop closure (e.g. Grice and Barry 1991). Our work in Newcastle revealed that 70% of audibly glottal(ised) segments were fully voiced, with the glottal percept being cued by creaky phonation (Foulkes and Docherty 2007: 61). 5.2 (T) 65 There are instances of creaky voice in Carlisle English that are perceived as glottal stops as well. The acoustic analysis of these instances would however exceed the scope of this study and thus these instances are counted towards the glottal stop variant. The aforementioned media attention given to glottal stops is most likely due to the stigmatisation it used to have and in (most) varieties still has, e.g. in the movie My Fair Lady (1964) the figure Eliza Doolittle, a poor girl from the East End of London, uses the glottal stop as one feature to identify her as Cockney speaker. In elocution lessons she learns how to ‘talk proper’ and to avoid the use of glottal stops. Wright’s comment about T-glottaling in Cumbria also al‐ ludes to a very negative perception of the glottal stop. Wells (1982: 324) com‐ ments that “it suffers some degree of overt stigmatization” and Milroy et al. (1994: 4) state that it is “one of the […] most stigmatized features of British English”. Coggle (1993: 42) emphasises the negative perception of intervocalic T-glottaling: Using the glottal stop between vowels is a bit like wearing a tattoo: whether you realise it or not, certain doors will be closed on you. It is a statement about you and about where you belong, or where you think you belong, in British society. A study conducted by Altendorf and Watt (2008: 29) shows that the use of glottal stops is not only observable in pupils at a comprehensive school in London, but that pupils at grammar schools and public schools now use this variant as well. From these results one can conclude that the use of this variant is no longer restricted to a certain social group and stigmatisation is declining. Even HRH William, Duke of Cambridge uses T-glottaling in word-final po‐ sition in the engagement interview (SkyNews 2010). The fact that these (young) representatives of the aristocracy now use this feature (though not in the most stigmatised intervocalic environment) reveals information about the status of T-glottaling. Word-final T-glottaling must have lost some of the stigmatization reported in earlier publications. Fabricius (2002: 134) discusses T-glottaling in RP and comes to the conclusion that it is now used in native-RP. Therefore, a formerly WC feature with covert prestige is now adopted even by upper-class speakers. While there is no discussion about the fact that T-glottaling is spreading across Britain, there are disputes about the origin of this feature. What is how‐ ever not disputed is the working-class origin. This stigmatisation seems to have eroded recently and T-glottaling has now become a ‘cool’ feature of young peo‐ ple’s speech. One indicator of this is the use of this variant in British Indie Rock 5 Sociolinguistic variables 66 music ( Jansen and Brato 2008) which makes it distinguishably British and sets it apart from the very influential music coming from the U.S. and Canada. 5.2.3 Data analysis In contrast to the vowel variables, data from the sentence list task and the reading task were included in the analysis for consonant variables in order to detect style shifts. An auditory analysis of the variable was conducted (cf. Foulkes, Docherty and Jones 2010). Tokens for (T) were collected from the in‐ terview starting from minute ten. In total 3,582 tokens were analysed, 861 from the text, 1,162 for the sentence list and 1,559 for interview style. Following Stuart-Smith (1999b: 182) four environments were distinguished: _#C (that comes), _#pause (right#), _#V (caught on), V_V (better). Two speakers did not read the text and the sentence list (6M38WC and 1F22WC) and for one speaker the interview data did not contain enough tokens (12M59WC). Therefore, the data available for these three speakers is restricted. The token but was not collected because of its very frequent, often unstressed occurrence. Assimilation of / t/ between word-final and word-initial position were omitted as well as consonant clusters, e.g. in past and negations such as isn’t or won’t. 5.2.4 Statistical modelling Similar to the vowel variables, a statistical model in Rbrul for the consonantal variables is used to identify predictors of language change. The analyses are based on the number of tokens and the variants are treated as dependent vari‐ able. The fixed social predictors are age, sex and social class as well as style and environment are treated as independent variables. The individual is run as a random effect while age is run as a continuous effect in the statistical model. For reasons of better representation, the figures which deal with age as social factor are split into three age groups: 22-39 years, 40-59 years and 60-78 years. 5.2 (T) 67 9 Fine-grained acoustic analysis of preaspirated / t/ would be desirable but is only possible to a certain extent for data recorded without an external microphone. 5.2.5 Results In this study five main phonetic variants for / t/ are identified: • a released stop [t], a released stop with heavy aspiration [tʰ] and a stop with preaspiration [ʰt]. For this analysis, all three allophones were col‐ lapsed into one variant 9 ; • a glottal stop [ʔ]; • an alveolar tap [ɾ] or a voiced alveolar stop [d], which was collapsed into one group; • an incomplete stop, which is realised as a fricative [ṱ]; • T-to-r. Only very infrequent is [ɹ], which can be found in Liverpool and Sheffield to a great extent, but also in other communities in the north of England. Linguistic constraints play an important role for this variant because it is restricted to the environment _#V (Broadbent 2008). In Carlisle English, this variant is used very infrequently. The data in Table 5.2.1 shows that the sample provides a sound basis for analysis and interpretation. There is a slight imbalance as the group of over 60-year-old speakers comprises about twice as many tokens as the other groups (34.7% vs 13.5%-18.8%). At this stage the distribution of [t] and [ʔ] for the five groups are visualised in Figure 5.2.1 and then the age cohorts are reduced to three groups: 22-39, 40-59 and 60-78. Table 5.2.1 provides the distribution of the data according to social class, sex and age. 5 Sociolinguistic variables 68 [t] [ʔ] [ɾ]/ [d] [ṱ] T-to-r Total Social Class MC N=987 50.4% N=742 37.9% N=64 3.3% N=164 8.4% N=3 0.2% N=1,960 54.7% WC N=872 53.8% N=568 35% N=48 3% N=127 7.8% N=7 0.4% N=1,622 45.3% Sex M N=988 55.8% N=521 29.4% N=92 5.2% N=165 9.3% N=6 0.3% N=1,772 49.5% F N=871 48.1% N=789 43.6% N=20 1.1% N=126 7% N=4 0.2% N=1,810 50.5% Age 22-29 N=196 28.3% N=448 64.6% N=11 1.6% N=37 5.3% N=1 0.1% N=693 19.3% 30-39 N=177 42.8% N=184 44.4% N=16 3.9% N=37 8.9% N=0 0% N=414 11.6% 40-49 N=219 47.4% N=187 40.5% N=7 1.5% N=46 10% N=3 0.6% N=462 12.9% 50-59 N=391 55.3% N=217 30.7% N=31 4.4% N=65 9.2% N=3 0.4% N=707 19.7% 60-69 N=672 65.4% N=236 23% N=34 3.3% N=82 8% N=3 0.3% N=1,027 28.7% 70-78 N=204 73.1% N=38 13.6% N=13 4.7% N=24 8.6% N=0 0% N=279 7.8% Total N=1,859 51.9% N=1,310 36.6% N=112 3.1% N=291 8.1% N=10 0.3% N=3,582 100% Table 5.2.1: Absolute numbers and percentages for each variant of (T) according to social class, sex and age In Figure 5.2.1 the apparent-time perspective reveals a striking cross-over effect between [t] and [ʔ] with each decade vastly outnumbering the other in the preference of glottal stops over alveolar stops. The oldest speakers use [t] in over 70% and [ʔ] in only 13% of all instances while the youngest age group sampled in this study almost turned these numbers around with 28% using [t] 5.2 (T) 69 and 65% using [ʔ]. In the following section these findings are discussed in more detail. Figure 5.2.1: Percentage of [t] and [ʔ] for (T) according to age The Rbrul results for / t/ are provided in Table 5.2.2 and 5.2.3. The results will be discussed below. Deviance: 4,008.846 df: 8 Grand mean: 0.519 Speaker ID random standard deviation: 0.559 Applications Value: Alveolar stop [t] Factors Log Odds Tokens (N) Propor‐ tion of Ap‐ plication Value Factor Weight (fw) Age p<0.01 Continuous 22-78 +1 0.047 Style p<0.01 Sentence list 0.768 1,162 0.671 0.683 Text 0.114 861 0.512 0.529 Interview -0.882 1,559 0.409 0.293 5 Sociolinguistic variables 70 Environ‐ ment p<0.01 V_V 1.197 666 0.737 0.768 _#pause -0.069 617 0.489 0.483 _#V -0.132 912 0.545 0.467 _#C -0.996 1,387 0.410 0.27 Table 5.2.2: Rbrul output results for the use of alveolar stops for all speakers Deviance: 3166.459 df: 9 Grand mean: 0.366 Speaker ID random standard deviation: 0.667 Applications Value: glottal stops [ʔ] Factors Log Odds Tokens (N) Propor‐ tion of Ap‐ plication Value Factor Weight Sex p<0.05 F 0.373 1,810 0.436 0.592 M -0.373 1,772 0.294 0.408 Age p<0.01 Continuous 22-78 +1 -0.066 Style p<0.01 Interview 1.238 1,559 0.461 0.775 Text -0.166 861 0.389 0.459 Sentence list -1.072 1,162 0.221 0.255 Environ‐ ment p<0.01 _#C 2.018 1,387 0.564 0.883 _#pause -0.074 617 0.342 0.482 _#V -0.267 912 0.250 0.434 V_V -1.676 666 0.134 0.158 Table 5.2.3: Rbrul output for the use of glottal stops for all speakers 5.2 (T) 71 10 t-to-r is very infrequent and is therefore not included here. 5.2.5.1 Social factors Age Younger speakers tend to produce more glottal stops and fewer voiceless al‐ veolar stops than older speakers in apparent time (see Figure 5.2.1 and Figure 5.2.2). Age is highly significant for the use of [t] and [ʔ] (p<0.01). This suggests that we are observing a change in progress where the standard variant [t] is replaced by [ʔ] (see Table 5.2.2 and Table 5.2.3). For the social factor age, the findings are in line with the findings across the county, e.g. Milton Keynes, Reading and Hull (Kerswill 2002: 208). Figure 5.2.2 provides the distribution of the variable according to the three previously distinguished age groups and for four of the five variants 10 I identified previously. One has to be careful, however, when interpreting these results. The increasing use of [ʔ] for / t/ by younger speakers could also be an indicator for age grading, i.e. young people prefer this variant over the more traditional form of [t] because it is more fashionable but when they get older, they return to using the more traditional form. As long as there is no evidence from a real-time point of view, both interpretations of the data need to be seen as possible. However, the use of glottal stops as variant is already quite established in the middle and old speaker group which, suggests that this distribution is not a result of age grading. Figure 5.2.2: Percentage of variants of (T) according to age 5 Sociolinguistic variables 72 The other two variants [ɾ] and [ṱ] found in this sample are not significant for age. Indeed, Figure 5.2.2 confirms that the age differences are minor. Sex Figure 5.2.3 reveals that women are leading the change towards [ʔ] which is statistically confirmed by the factor weight (fw: 0.576), i.e. females favour the use of glottal stops. Trudgill (1974) shows that in Norwich the use of glottal stops is a male norm. Similarly, Macaulay (1977 quoted in Stuart-Smith 1999b) reports that in Glasgow male speakers make more use of glottal stops than female speakers. Stuart-Smith (1999b) puts the results into perspective and states that while sex differences exist in certain social class groups these differences cannot be observed overall. On the other hand, studies conducted in Cardiff (Mees and Collins 1999), Tyneside (Milroy et al. 1994), Hull and Reading (Kerswill 2004) show that women are leading the change by favouring [ʔ] over [t] for / t/ . There‐ fore, Carlisle is another urban area where females use the incoming form [ʔ] more than male speakers. Figure 5.2.3: Percentage of variants of (T) according to sex Social class Figure 5.2.4 shows that larger differences between the two social classes do not exist. This is confirmed in the statistical analysis. A stratification according to social class is not observable while glottalisation has been reported to be highly stigmatised in various other studies (cf. Milroy et al. 1994; Williams and Kerswill 1999; Stuart-Smith 1999b). Stuart-Smith (1999b: 189) finds a “sharp divide” in the 5.2 (T) 73 distribution of T-glottaling across social class. She states that similar findings are made across Britain. Trudgill (1974: 96) mentions that “glottalisation […] of / t/ is inversely proportional to social class and social context.” Figure 5.2.4: Percentage of variants of (T) according to social class Figure 5.2.5 does not show clear stratification for the production of glottal stops between the different social groups with the exception of young MC males even though T-glottaling in intervocalic position is stigmatised in many community. Thus, glottal stops are not restricted and associated with only one social class group for the youngest cohort of speakers. If we take the apparent-time model for granted, it seems that WC male speakers (and to a lesser extent WC female speakers) were leading the change at an earlier point in time since the age group of 40-59 year-old WC male speakers uses this variant in about 20% of all cases while the other groups in this age group use the glottal stop considerably less. However, this difference in distribution neutralised across the generations with all groups now taking part in this change. Hence, there is the possibility that WC speakers were in fact leading the change at first but the glottal stop variant was then reinterpreted by the fol‐ lowing generation of speakers where covert prestige turned more into overt prestige. 5 Sociolinguistic variables 74 Figure 5.2.5: Percentage of [ʔ] in intervocalic position according to age group, sex and social class Bearing in mind that the use of glottal stops is highly stigmatised in intervocalic position, Figure 5.2.5 provides the data solely for glottal stops in this environ‐ ment. Both of the youngest female groups and the youngest WC male group use this variant to a similar extent while MC male speakers use it much less. The young MC speakers resist the incoming form in intervocalic position. It could be that the resistance to this change indexes young MC male identity. Style This section takes a closer look at the distribution of the variants across spoken styles. Table 5.2.4 presents a general overview of the distribution of variants according to style. 5.2 (T) 75 [t] [ʔ] [ɾ]/ [d] [ṱ] T-to-r Total Style Sentence list N=780 67.1% N=257 22.1% N=19 1.6% N=106 9.1% N=0 0% N=1,162 32.4% Text N=441 51.2% N=335 38.9% N=18 2.1% N=64 7.4% N=3 0.2% N=861 24.0% Interview N=638 40.9% N=718 46.1% N=75 4.8% N=121 7.8% N=7 0.4% N=1,559 43.5% Total N=1,859 51.3% N=1,310 37.6% N=112 3.0% N=291 7.9% N=10 0.3% N=3,582 100% Table 5.2.4: Absolute numbers and percentages for each variant of (T) according to style Once again, we can observe a crossover pattern (see Figure 5.2.6), this time be‐ tween sentence list, reading passage and interview style. The statistical results evince that [t] is favoured in sentence list style while [ʔ] is favoured in interview style. In general one can claim that the more formal the speech situation is, the higher the use of [t] is while [ʔ] is used in more informal situations, (i.e. here, in the sociolinguistic interview). In various studies this style shifting has been reported (Mees 1987; Trudgill 1988; Stuart-Smith 1999b). The shift in style for the incoming variant leaves us to conclude that the glottal stop is what (Labov 2001) coins as a marker. The speakers are aware of this change in progress and consciously make the decision to shift their pronunci‐ ation between different speech styles. Their use of glottal stops increases with the decrease of self-awareness, i.e. in sentence list style they use 23.1% of glottal stops, which increases to 47.2% in interview style. 5 Sociolinguistic variables 76 Figure 5.2.6: Percentage of variants of (T) according to style 5.2.5.2 Linguistic factors Looking next at linguistic effect, Table 5.2.5 below presents the overall results for the speech community. [t] [ʔ] [ɾ]/ [d] [ṱ] T-to-r Total Environment _#C N=569 41% N=782 56.4% N=7 0.5% N=29 2.1% N=0 0% N=1,387 37.8% _#V N=497 54.5% N=228 25% N=93 10.2% N=84 9.2% N=10 1.1% N=912 25.5% _#pause N=302 48.9% N=211 34.2% N=0 0% N=104 16.9% N=0 0% N=617 17.2% V_V N=491 73.7% N=89 13.4% N=12 1.8% N=74 11.1% N=0 0% N=666 18.6% Total N=1,859 51.9% N=1,310 36.6% N=112 3.1% N=291 8.1% N=10 0.3% N=3,582 100% Table 5.2.5: Absolute numbers and percentages for each variant of (T) according to environment 5.2 (T) 77 From the data in Table 5.2.7 it becomes apparent that there is a clear preference for different variants in a certain environment, which is affirmed by the Rbrul results (see Table 5.2.2 and Table 5.2.3). Figure 5.2.7 presents the percentile dis‐ tribution of variants according to environment with a decreasing order of [t]. The most striking finding is again the crossover pattern of [t] and [ʔ]. The _#C environment highly favours the use of glottal stops while in intervocalic position [t] is favoured. The statistical model ratifies this observation. With 0.882 the factor weight for glottal stops in _#C position is very high whereas [t] is dis‐ favoured in this position (fw: 0.264). These results are similar to Stuart-Smith’s (1999b) findings for Glasgow where a high percentage of [ʔ] in the _#C envi‐ ronment is attested. Reid (1978: 162) finds a similar distribution in Edinburgh. Figure 5.2.7: Percentage of variants of (T) according to environment A relatively low percentage of [ʔ] in V_V position is attested. With a centred weight of 0.153, this environment is disfavouring the use of glottal stops more than any other environment. As mentioned above, other researchers have com‐ mented that the use of a glottal stop in this position is more salient and stigma‐ tised than it is in word-final position. Stuart-Smith (1999b) describes similar results. The V_V environment is the one in which the use of glottal stops is lower than in other environment in Glasgow English. The results for the different environments also leave space to assume that there are internal linguistic constraints and that glottaling in the different en‐ vironments entered the community at different stages. This result is particularly apparent in the environments _#C and V_V. While the glottal stop is almost categorical in the former environment, in intervocalic position the use of [ʔ] 5 Sociolinguistic variables 78 reaches only 12.8%. This distribution, however, could also occur because of the slow adoption in different contexts due to the different levels of salience at‐ tached to the glottal stop in different environments. In her study on T-glottaling Stuart-Smith (1999b: 197) makes an important claim: “This qualitative analysis of T-glottaling suggests that […] not all glottals are the same. Linguistic attempts to shift socially up and down do not involve simply increasing or decreasing the number of glottal stops used […].” Indeed, this seems to be true for the Carlisle data as well. Even though social class does not play an immense role, the different phonetic environments are important when speakers choose between the different variants, especially [t] and [ʔ]. 5.2.6 Discussion The use of T-glottaling for the variable (T) is attested in Carlisle English. Word-final T-glottaling is a well-established feature in all age groups but it still increases in apparent time. In comparison, intervocalic glottaling, which is claimed to be the most stigmatised form in other varieties, is mainly found in speakers who are younger than 40 years old. For speakers over 40 this variant is very infrequent. Even though the reduction of an oral gesture from [t] à [ʔ] can be seen as phonetic weakening and therefore as language internal change, the use of glottal stops in this variety is clearly triggered by external motivation. Hickey (2012a) discusses the historical origin of a number of English language changes. For T-glottaling he comes to the conclusion: In London English the shift of / t/ is straight to / ʔ/ . In Dublin English, by contrast, / ʔ/ is part of a lenition chain which starts at / t/ and proceeds through / ṱ/ (an apico-alveolar fricative) to / h, r/ , then to / ʔ/ and possibly zero […]. Here the case for considering T-glottalling as an instance of lenition is more plausible. However, in both London English, Scottish English and Dublin English the shift can be regarded as stemming from an internal development in the sound system, this then spreading as the shift became a signal for vernacular metropolitan speech (at least in London and Dublin) and was thus emulated by other urban groups outside the capital (in the case of London). Just like Hickey, Docherty and Foulkes (1999: 68) see a system-internal explan‐ ation as a possible point of departure and they make an interesting observation, that “with pre-pausal [t] in particular, it might be argued that the stops have little information-bearing content, and the articulatory system may therefore be permitted a good deal of variability in the realisation of these stops.” This 5.2 (T) 79 11 Macaulay - Glasgow, Romaine - Edinburgh, Reid - Edinburgh, Macaulay - Ayr, Stuart-Smith - Glasgow. 12 They mention one environment where glottal replacement is even more common which is _C as in Gatwick. This was not an environment analysed in the present study. statement is particularly interesting in the light of the Carlisle English results. As was shown, prepausal glottaling is used by all speakers in the sample. How‐ ever, the quantitative distribution is by far not as extensive for _#pause as for the _#C environment. The result could hint at the higher variability in prepausal position, a point Docherty and Foulkes (1999) mention for this particular envi‐ ronment. Therefore, there is a good chance that language internal factors play an important role in this change. Another indicator that internal forces are at work is provided by Stuart-Smith (1999b: 192). In the aforementioned study of Glasgow English, she compiles data on the distribution of glottal stops in different environments from various Scot‐ tish communities in a comparable manner. 11 The results found in the present study are similar to her results for Glasgow English and the other Scottish Eng‐ lish varieties. Table 5.2.6 provides the frequency order of the interspeaker use of glottal stops in the different environments similar to Stuart-Smith’s (1999b) list. They are put on a scale from most frequent to least frequent. M most fre‐ quent _#C → _#pause → _#V → V_V least fre‐ quent F _#C → _#pause → _#V → V_V Table 5.2.6: Frequency order of [ʔ] for sex categories The distribution of glottal stops for men and women differ to some extent (see Table 5.2.1). However, the ranks are equally distributed for male and female speakers, i.e. both groups use [ʔ] most in _#C, and least in V_V. The frequency order for the other two environments differ as well but [ʔ] is more often used in _#pause position than in _#V position. In all cases V_V is the environment where glottaling is used least. In three out of five cases _#V is the environment which is used second to least. Altendorf and Watt (2008: 211) find for London that in _#C environment glottal replace‐ ment is more common than in the other environments 12 and in interview style / t/ is realised as [ʔ] in over 90% of all cases. Thus, the distribution of glottal stops across environments does not seem to vary much, even though the social dif‐ ferences between the varieties can differ dramatically. This could yet again hint at internal change. 5 Sociolinguistic variables 80 Nevertheless, internal factors cannot explain the entire change and in par‐ ticular the rapid spread across the country. Salience as a language external factor must be taken into account as well, in particular, for the use of glottal stops in intervocalic position. In this position the use of glottal stops remains stigmatised while in word-final preconsonantal position, glottal stops are almost used cat‐ egorically and are less salient. The use of glottal stops has become popular/ cool in many communities and has spread throughout Great Britain, also leaving its mark on Carlisle English. Young speakers in Carlisle use the glottal stop to identify with young urban British people. Originally spreading as a covert prestige feature in other vari‐ eties, in Carlisle English social class no longer plays an important role. Today, T-glottaling is indexing sex and age. Admittedly, nowadays the use of glottal stops is found in many communities across Britain. However, Carlisle is almost 500km away from London which could be one reason why the social markedness of this feature might not have travelled with the spread of this feature, i.e. a reinterpretation of the social con‐ notation of this variant must have taken place. The lack of differences in the social class distribution could also be due to the social structure of the city. It is in general a working class town and even though there are areas which are more affluent, there are no parts of the city which strictly can be defined as a MC or WC, e.g. Bellevue is an area which is described as a more affluent, but in this area I talked to two WC speakers working as cracker packers at McVitties. Almost all of my speakers have lived in different parts of the city. Thus, contact between people from different areas of the city and with different social class backgrounds is quite frequent. This is a completely different setting from e.g. Watts (2005) in Wilmslow or Milroy (1987) in Belfast where social classes lived in separate areas of the town/ city. The data analysis reveals that the glottal stop is not attested as a traditional feature in Carlisle English but rather as an innovation. Neither the data pre‐ sented here nor the historical literature search for Carlisle and/ or Cumberland provide evidence that the use of glottal stops is a feature of traditional Carlisle English. Thus, for now one must assume that the use of intervocalic glottal stops has diffused northward from the south as discussed by Kerswill (2003). 5.2.7 Conclusion The data presented in this part of the study shows that various variants of (T) - depending on the environment - are attested in Carlisle English. Most common however are the variants [t] and [ʔ] in all environments. Nevertheless, 5.2 (T) 81 13 Hickey claims that this is different for traditional Dublin English where we find a system-internal weakening. incomplete stops, which according to Docherty (2010) often go unnoticed in analyses, are attested as well. Yet tokens of this variant are distributed fairly evenly and they do not carry a lot of social meaning. Voiced alveolar stops and - very infrequently - T-to-r are attested in Carlisle English and complete the group of variants of / t/ . Some general trends are observable: the use of T-glottaling is increasing rap‐ idly in the given environments while the use of [t] is decreasing. The results are in line with Trudgill (1999a: 136): The glottaling of intervocalic and word-final / t/ is one of the most dramatic, wide‐ spread and rapid changes to have occurred in British English in recent times. […] It is spreading phonologically from more favoured to less favoured environments. With the help of quantitative methods I could show that the distribution of glottal stops differs according to the different environments in various contexts. External change seems to be a source for glottaling in Carlisle English. Nev‐ ertheless, internal constraints have been shown to be part this change as well. 13 Glottaling has become a supralocal change but if one looks at the distribution of [ʔ] in different environments across varieties, there is a certain order of ac‐ quisition, which suggests that internal change is also part of the story. This chapter has added to the discussion about T-glottaling in the north-west of England. Contrary to other varieties, in Carlisle English social class is not a crucial external factor for the use of this variant. This finding is quite unique in the urban British sociolinguistic landscape. Previously, researchers have often connected WC speech with this variant while it was claimed that MC speakers were less likely to use this variant. In fact, in Carlisle English speech glottal replacement must have lost the covert prestige which it obviously had in other communities. This is most likely due to the social situation in the city. 5.3 (R) The third sociolinguistic variable I examine is the variation of (R) in non-post‐ vocalic position in Carlisle English which shows some complex sociolinguistic variation which is worth studying. 5 Sociolinguistic variables 82 14 Not only the use of labiodental / r/ but also TH-fronting and L-vocalisation are found in language acquisition processes when children learn English. 5.3.1 Historical and geographical background Different realisations of / r/ exist in the north of England. A widespread (and in most cases traditional) realisation of / r/ is as tap [ɾ] (Cruttenden 2008: 88). In the SED data, Wells (1982: 370) finds alveolar trills for the north of Cumbria and indeed, in older recordings of Carlisle English speakers trills are attested in low numbers. Jansen and Langstrof (in preparation) also find trills in West Cumbria. Wells (1982: 368) states that “an alveolar tap, [ɾ], seems quite widespread in the north of England as a rival to the usual post-alveolar approximant, [ɹ].” Stuart-Smith (2008: 64f) reports alveolar taps and approximants for Scottish English and Scots, the latter group favouring the tap realisation. For Lancashire and Cheshire, Beal (2008: 140) mentions the use of a flap. Llamas (1998: 107f) points out the infrequent usage of taps [ɾ] in Middlesbrough, but focuses on the variation between [ɹ] and [ʋ] while Jansen (submitted) still finds a high use of taps in Maryport English, a West Cumbrian community. The use of taps is de‐ creasing across the whole of the north of England (e.g. Watt and Milroy 1999; Llamas 2001b; Marsden 2006) and it is being replaced by a voiced postalveolar approximant [ɹ]. This change corresponds to what Kerswill (2003) sees as a countrywide levelling process. In Carlisle English, the tap is the traditional norm. This chapter aims to investigate the levelling process of prevocalic / r/ in the community. Another variant of (R) is receiving a lot of attention at the moment: the re‐ placement of the approximant [ɹ] with the labiodental [ʋ] for / r/ (Kerswill 2003: 231). Various sociolinguists have commented on the spread and use of it in speech communities (e.g. Torgersen 1997; Foulkes and Docherty 2000; Marsden 2006). Torgersen (1997: 40) analyses the fronting of (R) from approximant to labio‐ dental position in the counties surrounding London. He reports that 11.1% of all tokens in his data are realised as [ʋ] and stresses that the majority (84.1%) of this realisation were produced by young female speakers (under 20). Three of his young female informants were even near-categorical in their usage of [ʋ]. In their report on the use of [ʋ] in Derby and Newcastle, Foulkes and Docherty (2000) use dialectological, sociolinguistic and acoustic evidence to reconstruct the history of the labiodental variant. They suggest two possible sources for the use of [ʋ]: the use of labiodentals was for a long time equated with a speech defect or affectation. “The conception of [ʋ] as an infantilism 14 or pronunciation problem has filtered down relatively undiluted over the years. […] Several au‐ 5.3 (R) 83 thors have also identified [ʋ] as a feature of ‘affected’ upper class speech” (Foulkes and Docherty 2000: 31). An analysis of historical sources provides evi‐ dence that labiodental / r/ was not a feature of Cockney English in the 19 th century. Their tentative finding is that the introduction of [ʋ] originated in the East End of London, possibly due to a language contact scenario. The large Jewish community that arrived in London between 1880 and 1900 could have been the innovators of this change as they tried to modify their native uvular [ʁ] towards the English norm of alveolar approximant [ɹ]. The labiodental [ʋ] then started to spread beyond the London East End and the feature is now found in varieties across England. More evidence however is yet needed for this ten‐ tative explanation. Beal (2007) doubts that labiodental sounds were not origi‐ nally part of the Cockney English inventory. She provides an example from the ‘how to’ manual Enquire Within upon Everything (1878; quoted in Beal 2007: 41) and comes to the conclusion “that labio-dental / r/ was a salient feature of London English in the late nineteenth century, but was associated with the lower middle-class rather than the working-class Cockneys.” The example that leads Beal to this conclusion also provides us with some information about the per‐ ception of the labiodental variant at the time. Beal comments that one part deals with “proper pronunciation” which “set out to warn him, or more likely her, against the worst social gaffes, in speeches as well as behaviour” (Enquire Within 1878: 54 quoted in Beal 2007: 42). Thus, labiodental / r/ is indeed commented on as a Cockney variant and is evaluated rather negatively in London speech at the end of the 19 th century. The general perception of [ʋ] is still negative. Comments about the use of [ʋ] in recent publications do not really confirm Trudgill’s (1988: 40) statement that “[…] when a speech defect spreads to a majority of the population it is clearly no longer to be regarded as such.” This claim is to some extent ambiguous. It is not clear from the extract who decides what is acceptable as a linguistic variant and what is not. Is Trudgill talking about a single speaker or a speech com‐ munity? If this is so, then this black and white distinction is far from reality. For example, the statement by two young male speakers from the Middlesbrough data set Llamas (2001: 136) talking about a person “who couldn’t say her r’s,” shows that there is a time overlap of negative commenting on what speakers think is a speech defect and their own use of it which is by all means unconscious. Thus, at least in Middlesbrough one can classify the infiltration of [ʋ] as a non-salient dialect feature. Even though young speakers of various varieties actively use labiodental [ʋ], they still negatively evaluate it, e.g. as a speech defect. These speakers are not aware of their own active use of this variant. 5 Sociolinguistic variables 84 15 Arguably, low frequencies of a form could be highly salient but could go unnoticed as well. I would argue that it depends on the nature of the variant and the community’s perception of it, in how far such a low-frequency form is consciously perceived by interlocutors. The processes which lead to the ‘recycling’ of a variant are yet to be observed. Similar to the re-emergence of the use of glottal stops, Beal (2007) also argues that the labiodental / r/ was apparent in the north-east of England from the late 18 th century onwards. Hence, the use of this feature cannot be called ‘innovation’ in this area. Although the apparent-time evidence cited by Foulkes/ Docherty does suggest that the recent spread of this feature is due to dialect levelling, historical sources point to the possibility that, in Newcastle at least, there was a local model already present: a la‐ biodental / r/ derived from the ‘burr’ which had been one of the variants heard in Newcastle from the late eighteenth century onwards (Beal 2007: 43). Therefore, in Newcastle English, the labiodental variant was heard before the diffusional wave of labiodental / r/ arrived in the north-east. Hickey (2002) talks of an ‘ebb and flow’ movement of variation in a community. This means that a form is apparent in a community with a varying level of quantitative represen‐ tation. This also seems to be the case for labiodental / r/ in Newcastle. 15 5.3.2 Data analysis Similar to the data analysis in §5.2, I analyse tokens for (R) in four environments for three different styles: interview, reading passage and sentence list style. Fol‐ lowing Watt, Llamas and Johnson (2010: 275) an auditory analysis of tokens from four different environments is carried out: V#_V (as in a road), #C_V (as in crown), V_V (as in berry), and V_#V (as in far away). However, not the entire sample could be used in this part of the study. For 12M59WC, 10F49WC and 19F68MC only the tokens from the reading tasks were collected because the interviews did not provide enough tokens. 5.3.3 Statistical modelling Four logistic regressions were carried out with [ɹ], [ɾ], [ʋ] and Ø as dependent variables (the variants are the application value) and with individual speaker as a random effect and age as continuous effect. The results are provided in Table 5.3.1-5.3.5. 5.3 (R) 85 16 Foulkes (1997: 78) mentions that in some cases glottal stops are inserted instead of linking / r/ which is also the case in Carlisle English. This possible realisation is also counted towards the zero form. 5.3.4 Results The data representation in this chapter is similar to the one in §5.2 for (T). Overall, 3,147 tokens for (R) were analysed and four main phonetic variants were identified: • a voiced postalveolar approximant [ɹ]. This is the more standard variant and is now found in almost all varieties of English in England; • an alveolar tap [ɾ] or trill [r]: the latter is the more traditional form and is used only in a handful of occasions and therefore both allophones are counted towards the tap variant; • zero (non / r/ ) 16 realisation: in V_#V position linking / r/ is on the decline. Thus, a hiatus (which is sometimes filled) between the two vowels is ob‐ served; • labiodental [ʋ]: this variant is a fairly recent innovation in Carlisle Eng‐ lish. Along with T-glottaling and L-vocalisation, it is another consonantal feature that is spreading across the country (e.g. Kerswill 2003). [ɹ] [ɾ] Ø [ʋ] Total Social Class MC N=1,219 70.9% N=323 18.8% N=79 4.6% N=99 5.8% N=1,720 54.7% WC N=792 55.5% N=535 37.5% N=46 3.2% N=54 3.9% N=1,427 45.3% Sex M N=923 58.2% N=543 34.2% N=57 3.6% N=64 4.0% N=1,587 50.4% F N=1,088 69.7% N=315 20.2% N=68 4.4% N=89 5.7% N=1,560 49.6% Age 22-29 N=407 64.7% N=65 10.3% N=38 6% N=119 18.9% N=629 20% 30-39 N=260 69% N=81 21.5% N=22 5.8% N=14 3.7% N=377 12% 5 Sociolinguistic variables 86 17 Even though this change has progressed quite a bit, sociolinguistic patterns are still observable. 40-49 N=276 72.4% N=86 22.6% N=14 3.7% N=5 1.3% N=381 12.1% 50-59 N=428 67.7% N=168 26.6% N=23 3.6% N=13 2.1% N=632 20% 60-69 N=488 56.7% N=349 40.5% N=22 2.6% N=2 0.2% N=861 27.4% 70-78 N=152 56.9% N=109 40.8% N=6 2.2% N=0 0% N=267 8.5% Total N=2,011 63.9% N=858 27.3% N=125 4% N=153 4.9% N=3,147 100% Table 5.3.1: Absolute numbers and percentages for each variant of (R) according to social class, sex and age Table 5.3.1 provides the overview of the data for (R) for the social factors social class, sex and age. In the Carlisle data all four variants are found for both male and female and MC and WC speakers although [ɹ] is the preferred realisation of (R) for all speaker groups across apparent time. In Table 5.3.1 the results for six age groups are presented. In the following they are cut down to three age groups which provides a better overview of the data. Overall, in Table 5.3.1 there is a change in progress with two main variants ([ɹ] and [ɾ]) 17 and two innovative forms (Ø and [ʋ]). Table 5.3.2-5.3.5 present the statistical results for the four dependent variants. In the following the external factors age, sex, social class and style as well as the internal factor environment are discussed in detail in order to detect external and internal constraints. Deviance: 3,020.099 df: 9 Grand mean: 0.639 Speaker ID random standard deviation: 0.901 Applications Value: Factors Log Odds Tokens (N) Proportion of Applica‐ tion Value Factor Weight 5.3 (R) 87 approximant [ɹ] Sex p<0.05 F 0.378 1,560 0.697 0.593 M -0.378 1,587 0.582 0.407 Social Class p<0.01 MC 0.489 1,720 0.709 0.62 WC -0.489 1,427 0.555 0.38 Style p<0.05 Interview 0.187 1,470 0.598 0.547 Text -0.090 672 0.637 0.478 Sentence list -0.098 1,005 0.700 0.476 Environment p<0.01 #C_V 1.835 1,27 0.872 0.862 V#_V 0.311 594 0.660 0.577 V_V -0.822 706 0.435 0.305 V_#V -1.324 570 0.347 0.21 Table 5.3.2: Rbrul output for use of approximants for all speakers Deviance: 2,413.3 df: 10 Grand mean: 0.273 Speaker ID random standard deviation: 1.323 Applications Value: tap [ɾ] Factors Log Odds Tokens (N) Proportion of Applica‐ tion Value Factor Weight Age p<0.01 Continuous scale 22-78 +1 0.05 3,147 Sex p<0.01 M 0.614 1,587 0.342 0.649 F -0.614 1,560 0.202 0.351 Social Class p<0.01 WC 0.854 1,427 0.375 0.701 MC -0.854 1,720 0.188 0.299 Style p<0.01 Sentence list 0.199 1,005 0.225 0.55 Interview 0.190 1,470 0.327 0.547 5 Sociolinguistic variables 88 Text -0.389 672 0.226 0.404 Environment p<0.01 V_V 1.474 706 0.520 0.814 V_#V 0.893 570 0.432 0.709 V#_V -0.657 594 0.212 0.341 #C_V -1.710 1,277 0.093 0.153 Table 5.3.3: Rbrul output for use of taps for all speakers Deviance: 770.144 df: 8 Grand mean: 0.049 Speaker ID random standard deviation: 1.201 Applications Value: labio-dental [ʋ] Factors Log Odds Tokens (N) Proportion of Applica‐ tion Value Factor Weight Age p<0.01 Continuous scale 22-78 +1 --0.117 3,147 Style p<0.01 Text 0.322 672 0.046 0.58 Sentence list 0.253 1,005 0.061 0.563 Interview -0.575 1,470 0.041 0.36 Environment p<0.01 V#_V 2.223 594 0.128 0.902 V_V 0.581 706 0.045 0.641 #C_V 0.073 1,277 0.034 0.518 V_#V -2.876 570 0.002 0.053 Table 5.3.4: Rbrul output for use of labiodental / r/ for all speakers Deviance: 494.808 df: 8 Grand mean: 0.04 Speaker ID random standard deviation: 0.721 5.3 (R) 89 Applications Value: zero Ø Factors Log Odds Tokens (N) Proportion of Applica‐ tion Value Factor Weight Age p<0.01 Continuous scale 22-78 +1 -0.033 3,147 Style p<0.01 Text 1.560 672 0.091 0.826 Sentence list -0.719 1,005 0.014 0.328 Interview -0.841 1,470 0.034 0.301 Table 5.3.5: Rbrul output for use of zero for all speakers 5.3.4.1 Social factors Age Figure 5.3.1 provides the distribution of the four variants of / r/ in apparent time. The ubiquitous levelling pattern of the local form [ɾ] and the concomitant increase of the supraregional form [ɹ] partly applies for (R) in Carlisle English as the Figure 5.3.1 below reveals. Yet age is not a predictor of the use of the standard form [ɹ]. However, age is highly significant for the realisation of taps. A change in progress is the decrease of taps across apparent time (p<0.01). The oldest age group still produces [ɾ] in about 40% of all cases while the production of the tap variant decreases to about 15% for the youngest age group. The comments about [ɾ] in the volume on Urban Voices in other Northern cities (e.g. Sheffield (Stod‐ dard, Upton and Widdowson 1999); Newcastle (Watt and Milroy 1999); (Llamas 2001b for Middlesbrough) hint at a very weak representation of this traditional variant in the samples from the different varieties while the frequency of taps in peripheral northern areas is still comparatively high ( Jansen submitted). Even though the tap variant is rapidly decreasing in Carlisle English, there is still a fair amount of use observed in this variety. As is demonstrated below, internal constrains play a role for the distribution of this variant. 5 Sociolinguistic variables 90 Figure 5.3.1: Percentage of variants of (R) according to age As a third variant for / r/ , the labiodental variant [ʋ] makes its way inroads into the variety and age is highly significant for this variant (p<0.01). Less than 2% of the middle age group uses this variant before it rockets upward to about 15% in the youngest age group. Thus, a similar distribution of the traditional form [ɾ] and the innovative form [ʋ] is found for the youngest group in the sample yet with different patterns in apparent time. The fourth variant that is also increasing in apparent time is the zero form (p<0.01). In relation to the other variants, the relative distribution is quite low which derives from the fact that this variant is restricted to the V_#V environ‐ ment. In the section on environmental differences this distribution is discussed in more detail. Overall, if the variants are divided into winners and losers in this sample, labiodental [ʋ] and the zero form in linking / r/ position are definite winners with their increase in apparent time, while the local tap realisation must be counted as the main loser because of its losses in apparent time. Nevertheless, even though the position of [ɹ] is stabilised, a slight decrease of the variant between the group of the young and middle aged speakers is observable and as discussed before, age is not significant for the distribution on [ɹ]. In the youngest age cohort, the standard variant is fairly stable while the innovative variants increase at the expense of the traditional tap variant. The diffusion of the variant [ʋ] arrived earlier in Newcastle English (Foulkes and Docherty 2000) than in Carlisle English. They base their analysis on data 5.3 (R) 91 recorded between 1994 and 1996 and they state that none of the speakers who are older than 45 years use this feature. Figure 5.3.1 presents the data for the variety. In the 40-59 age group, 18 tokens of the labiodental variant are recorded. A closer look reveals that eight of these tokens are attested for 12F56WC, a speaker who is in frequent contact with university students and people from outside of Carlisle. There are also ten tokens that are produced by other speakers. However, one must assume that these single instances per speaker are not really part of the language change process since the change only takes off for the group of 40-49 year old speakers. Therefore, it can be assumed that there is a time lag of about ten years between the arrival of the diffusional wave of [ʋ] in Newcastle English and in Carlisle English. All in all, a comparison of the present data on the use of the innovative form [ʋ] with the use of it in other varieties reveals that it is definitely not (yet) as established as in the south-east (cf. Trudgill 1988; Torgersen 1997). Sex Sex is a significant social factor for the variants [ɹ] (p<0.05) and [ɾ] (p<0.01). Figure 5.3.2 confirms the change away from the local non-standard variant in particular for female speakers. Women favour [ɹ] (fw: 0.591) while men favour [ɾ] (fw: 0.643). However, the distribution of Ø and [ʋ] for male and female speakers does not differ significantly and thus sex does not predict the distri‐ bution of Ø and [ʋ]. Figure 5.3.2 splits up the sex groups into the three age groups in order to obtain more information on the distribution of the four variants. In the group of 60-78-year old female speakers, the percentage of [ɾ] is still 37.4% but it de‐ creases steadily to only 5.5% for the youngest female speakers. The use of the traditional form also decreases in apparent time for the male speakers but the decrease is less dramatic than for the female speakers. Men use the standard form [ɹ] significantly less than for female speakers. 5 Sociolinguistic variables 92 Figure 5.3.2: Percentage of variants of (R) according to sex The distribution of [ɹ] and [ɾ] reflects Labov (2001) predictions: the female speakers here conform more to norms that have overt prestige, in this case [ɹ], while the male speakers in this sample maintain [ɾ] which is the local Carlisle English norm. Indeed, in all three age groups presented in Figure 5.3.3, the use of approximants is higher for women than for men. The incoming variant [ʋ] is equally distributed among the two sex groups in the youngest age cohort. Sex is therefore not a predictor for this variant. This is a rather unusual finding and I further discuss this issue below in the section on social class. The third incoming form zero is attested for both sex groups in all age cohorts. The increase seems to progress fairly slowly and sex is not a predictor for the use of this form either. 5.3 (R) 93 Figure 5.3.3: Percentage of variants of (R) according to age and sex Social class Figure 5.3.4 provides a detailed view on the distribution of variants for / r/ ac‐ cording to social class. Overall, the importance of social class as a factor influ‐ encing the use of variation of prevocalic / r/ is only limited. Social class is only a highly significant factor (p<0.01) for [ɹ] and [ɾ] while it is not a predictor for the innovative forms [ʋ] and Ø. Figure 5.3.4: Percentage of variants of (R) according to social class 5 Sociolinguistic variables 94 MC speakers are generally predicted to use the standard form more than WC speakers. This is also the case in this community. MC speakers use the approx‐ imant as standard form more often than WC speaker. At the same time, WC speakers adhere more to the local non-standard norm. The lack of significance of social class as predictor for change in the two innovative variants is unusual because in other varieties the distribution of these variants is predicted by social class. Thus, in the following a closer look at the distribution of (R) according to age, sex and social class is provided. The data presented in Figure 5.3.5 provides more insight in the distribution of this variable. The group of WC male speakers and MC female speakers in the 22-39 year old age cohort use the labiodental variant to almost equal share. In the light of Llamas’s (Llamas 1998; Llamas 2001b: 130f) finding, this result is striking. In her data, predominantly young WC male speakers use [ʋ]. Similarly, Foulkes and Docherty (2000) find that WC speakers from Derby use the labio‐ dental variant more often than MC speakers. However, in Newcastle English, the use of [ʋ] is prevalent in young MC female speakers. While in these varieties only one group is leading the change, in this sample both groups have leading roles in the distribution of labiodental / r/ . A closer look at the single speakers in these two groups reveals that a divide in the WC male speaker group exists. 1M22WC’s use of / r/ stands out. He belongs to the group of 22-39 year old WC males and uses the innovative labiodental form in 31% of all instances. On the other hand, 5M36WC uses [ʋ] in 2.4% of all instances and 6M38WC in 7.7%. Figure 5.3.5: Percentage of variants of (R) according to age, sex and social class 5.3 (R) 95 For the MC female group one might be seeing the effect of dialect contact with other cities. Three MC female speakers in the youngest age cohort studied in Edinburgh, Durham and Newcastle. The contact with speakers from these cities could have led to the increased adoption of labiodental / r/ because e.g. in New‐ castle, this is a feature young MC females adopt. However, 2F23MC - who did not go to university in a different city - still uses this variant in 21.4% of all instances in the MC cohort. The results for the 22-39 year old MC female speakers are not as distinct as expected, making it more difficult to pinpoint which group is leading the change. Due to style differences, it seems that it is a change from above which makes is most likely that the MC female speakers are in fact responsible for this change and leading it. Hence, one has to ask why one WC speaker is more or less an outlier. The sociolinguistic background of this speaker can provide an answer here. 1M22WC’s extensive use of [ʋ] might be the result of working as a (warehouse) clerk at a British retailer. Even though he is not in direct contact with customers, he is in contact with the predominantly female salesclerks who in many cases represent (L)MC speakers, a group that is likely to use incoming overt presti‐ gious forms more than any other group. Hence, one needs to be aware of the effect that an individual’s occupation has on language choice within a social class. One can also claim that despite the statistical insignificance of sex and social class, MC female speakers are most likely the group to lead the change towards [ʋ]. The indifference between male and female speakers is probably due to the WC male outlier discussed above. Social class This section will have a closer look at the distribution of the variants across spoken styles. Table 5.3.6 presents a general overview of the distribution of var‐ iants according to style. 5 Sociolinguistic variables 96 [ɹ] [ɾ] Ø [ʋ] Total Style Sentence list N=704 70.1% N=226 22.5% N=14 1.4% N=61 6.1% N=1,005 31.9% Text N=428 63.7% N=152 22.6% N=61 9.1% N=31 4.6% N=672 21.4% Interview N=879 59.8% N=480 32.7% N=50 3.4% N=61 4.2% N=1,470 46.7% Total N=2,011 63.9% N=858 27.3% N=125 4% N=153 4.8% N=3,147 100% Table 5.3.6: Absolute numbers and percentages for each variant of (R) according to style In all three styles the standard variant [ɹ] is preferred over the other options. However, in the less formal speech style, i.e. interview style, the local non-standard variant [ɾ] is used in 32.4% of all cases while [ɹ] is used in 59.9%. Even though the traditional form is still used in almost one third of all cases in interview style, this realisation only plays a minor role when it comes to dif‐ ferent speech styles. Figure 5.3.6 provides the information that the most obvious style change is the increase of the standard variant with increasing formality. Due to this result one can assume that this shift towards the standard variant is conscious. It seems that the decrease of the local norm from the less formal to more formal style is also a conscious choice the speakers make even though a difference between text and sentence list style does not exist. 5.3 (R) 97 Figure 5.3.6: Percentage of variants of (R) according to style The distribution of labiodental [ʋ] across the different styles should be men‐ tioned here as well. The use of labiodental / r/ increases with the formality of the style. As discussed above, speakers in Middlesbrough are not aware of their own use of [ʋ] even though they notice it in other people. The results here suggest that the speakers are aware of this variant to some extent and associate it with formality more than with a speech defect. Nevertheless, none of the speakers comment on this variant. The distribution of Ø for the different styles is puzzling. There is no clear line of argumentation for this variant. The small peak for Ø in text style might be explained by the nature of the text. It could be that certain phrases triggered the omission. In fact, linguists have argued that linking / r/ (as well as intrusive / r/ ) is more common in connected speech than in formal speech situations. Never‐ theless, one cannot observe such a style pattern for this variant. For now, one can only comment on the lack of a pattern. A possible future research objective would be to go back and find other means of style variation (or use a different reading passage in order to find out whether the passage is skewing the results), which could give us more insight into the variation of (R) in word-final (V_#V) position. 5.3.4.2 Linguistic factors This part focuses on possible internal constraints for the distribution of the var‐ iants for / r/ . Table 5.3.7 provides an overview on the data which is split up according to the four environments #C_V, V#_V, V_#V and V_V. 5 Sociolinguistic variables 98 [ɹ] [ɾ] Ø [ʋ] Total Environment #C_V (cluster) N=1,114 87.2% N=119 9.3% N=0 0% N=44 3.45% N=1,277 40.6% V#_V (initial) N=392 65.99% N=126 21.2% N=0 0% N=76 12.79% N=594 18.9% V_#V (final) N=198 34.7% N=246 43.1% N=125 21.9% N=1 0.2% N=570 18.1% V_V (intervocalic) N=307 43.5% N=367 52% N=0 0% N=32 4.5% N=706 22.4% Total N=2,011 63.9% N=858 27.3% N=125 4% N=153 4.9% N=3,147 100% Table 5.3.7: Absolute numbers and percentages for each variant of (R) according to environment The data reveals that linguistic constraints play an important role in the reali‐ sation of (R) in Carlisle English. Compared to the external factors, the internal factor environment is statistically significant for every variant. Figure 5.3.7 dis‐ plays the overall distribution of variation across the different environments. Figure 5.3.7: Percentage of variants of (R) according to environment 5.3 (R) 99 Even though the approximant is used almost categorically in #C_V position and for roughly 70% of all cases in V#_V position, in V_#V and V_V position the local variant [ɾ] is still very common and used more than any other variant. As dis‐ cussed before, the use of the zero form is restricted to the V_#V environment. The innovative form [ʋ] is attested for the three environments #C_V, V#_V and V_V, however, with differing distributions. For V_#V this variant is not attested. In the #C_V environment the standard variant [ɹ] is favoured (fw: 0.864) while taps are strongly disfavoured (fw: 0.152) and to some extent the labiodental variant [ʋ] is favoured in this environment (fw: 0.514), however, by no means as much as the approximant. Not only does the position of / r/ play a role for the production of [ɾ] but also the preceding segment seems to be important. Wells (1982: 368) makes some tentative comments about environments that could favour a tap realisation: An alveolar tap, [ɾ], seems quite widespread in the north of England as a rival to the usual post-alveolar approximant, [ɹ]. The conditions favouring the use of [ɾ] are not exactly known; I have the impression that the environments in which it occurs most readily are that of a preceding labial (pray, bright, frog), a preceding dental (three), and intervocalically (very, sorry, pair of shoes). Nor is its geographical spread known. Figure 5.3.7 already showed that in fact the use of [ɾ] is still preferred over the use of [ɹ] in V_#V and V_V position, conditions that Wells (1982) also mentions. In order to investigate which consonants would favour the tap in an initial con‐ sonant cluster, I extracted all tokens which had a tap realisation in #C_V position and looked at the preceding phonological environments. The results are pro‐ vided in Figure 5.3.8. The voiced bilabial stop triggers the production of [ɾ] more than any other preceding consonant which is similar to Wells’s observations. In addition, voiced stops seem to be favoured in preceding position if it is not an alveolar stop. Hickey (2004: 38) points out that a preceding alveolar stop is less likely to be followed by a tap due to articulatory complexity. However, in Irish English al‐ veolar stops are dentalised which makes it possible to use taps after these stops. In Carlisle English dental sounds do not exist and as Figure 5.3.8 suggests, taps are avoided after alveolar stops. However, in the traditional Edenside dialect, dental sounds are observed. Ellis (1889: 543) writes that “after t, d the r is invar‐ iably dental.” Wright (1978: 14) also mentions that dental sounds are produced before taps. “Phonetically speaking we have [...] some dental t’s so that e.g. ‘tractor’ sounds like ‘thractor’.” 5 Sociolinguistic variables 100 Figure 5.3.8: Preceding environment in #C_V position when [ɾ] is used (N = 119) Hence, the dental / t/ was probably part of the Carlisle English inventory but it was levelled and replaced by approximants. One also can conclude that in Carlisle English, taps are foremost produced in intervocalic position but the use of them is very restricted if / r/ is preceded by a consonant. Therefore, we can assume that this internal constraint has probably influenced the loss of [ɾ] in #C_V position. For V#_V the picture is quite complex in that the standard form [ɹ] (fw: 0.571) and the innovative labiodental [ʋ] (fw: 0.909) are favoured in this environment while taps are disfavoured (fw: 0.336). Thus, word-initial tap production is less likely while the new form [ʋ] is very likely to be produced in word-initial po‐ sition. From the distribution of the approximants among the environments, it seems that labiodental / r/ occupied this environment first before it spread to the other environments. In V_V position the local form [ɾ] is still well represented for Carlisle English speakers and indeed this environment favours the use of taps (fw: 0.811). La‐ biodental sounds are also favoured in this environment (fw: 0.644) but approx‐ imants are disfavoured (fw: 0.311). This result confirms Wells’s (1982: 368) as‐ sumption that taps are still fairly common in intervocalic position. Overall, the results provided in this subchapter clearly project very strong internal constraints for / r/ . Environment as well as preceding consonantal sounds as internal factors influence the choice of variant in Carlisle English. On the one hand, the incoming variant [ʋ] is preferred in word-initial position (V#_V). Torgersen (1997: 43) finds a similar distribution for labiodental / r/ in the south of England. On the other hand, in #C_V or cluster position stop sounds that are realised at the alveolar ridge inhibit a tap realisation. Even though there 5.3 (R) 101 are records that / t d/ were (partially) dental at the end of the 19 th century as Ellis (1889) shows, dental sounds are hardly part of the Carlisle English inventory anymore. Today, (R) following alveolar stops are not produced as taps anymore but almost categorically as alveolar approximants. The change from tap to ap‐ proximant realisation is not documented and therefore one can only speculate how this change took place. It is, however, more likely that dental stops were replaced by alveolar stops which then had the knock-on effect that approximants replaced taps in CrV position in order to avoid the articulatory complexity men‐ tioned above. 5.3.5 Discussion The auditory analysis of (R) holds some complex results. The observation that the local form is replaced by a more standard or supralocal form is generally no surprise. However, Carlisle English has passed that stage. The distribution of [ɹ] is already stable because age is not a predictor for change for this variant. But a steady decline of [ɾ] is observed. In addition, two innovations are observed in the use of prevocalic / r/ : labiodental [ʋ] and the zero form in linking / r/ position. The analysis of the data in the previous section suggests that three geolin‐ guistic processes have to be distinguished here: the replacement of [ɾ] with [ɹ] must be seen as a levelling process where the local marked variant is replaced by a more supraregional variant as was seen in Figure 5.3.1. This process is now on the whole completed in the community. As mentioned before, levelling pro‐ cesses are under way in basically all varieties of English in England which still have a traditional / r/ variant apart from the approximant. The changes in apparent time suggest that [ʋ] is to a certain extent replacing the remains of the local norm [ɾ] and the standard form [ɹ]. The northward diffusion of [ʋ] as already observed by others (e.g. Marsden 2006) continues. The data suggests a more recent adoption of this change in Carlisle English than in other areas further south. The increase of this variant is foreseeable. This is also observable for many other varieties in recent years. It is striking that statistically neither sex nor social class is a predictor for the use of this incoming variant. A situation which stands in contradistinction to the results found in Derby (Foulkes and Docherty 2000) and Middlesbrough (Llamas 2001b). However, as explained above, this possibly has to do with the work environment of one of the WC speakers who uses quite a lot of labiodental / r/ . If the data of this speaker is omitted, young MC female speakers clearly lead the change towards the overt norm. The results for the use of [ʋ] among the WC male group is heavily influ‐ enced by the fact that one of these speakers works as sales and warehouse clerk. 5 Sociolinguistic variables 102 Thus, the results of the sociolinguistic patterns found for Carlisle English are similar to Newcastle English (Foulkes and Docherty 2000) where young MC females lead the change. Another point of interest is the underlying processes of this diffusional change, e.g. why do certain features diffuse and what are the prerequisites for the diffusion of a feature. Llamas (2001b: 136) picks up these points and explains that […] it seems, […], that the degree of salience and the covert prestige carried by a variant are not necessarily predictors of which innovations will be diffused, as a feature with seemingly low salience and no covert prestige, such as [ʋ], can be diffused as quickly as a feature with high salience, such as (th) fronting. The processes underlying the initial spread of [ʋ] across England are indeed different from the ones for T-glottaling or TH-fronting. However, the result, i.e. the diffusion of this feature across the country is similar to the other features. The results for Carlisle English support Llamas’s (2001b) observations about low salience and no covert prestige for [ʋ]. Because of the observed style shift of labiodental / r/ one has to assume that the speakers are aware of the change. This change from above is also in line with the findings in the sample. The use of labiodental / r/ must be interpreted as carrying overt prestige in Carlisle English. In order to better understand the different ways of diffusion, perceptual testing would be one way to further increase our understanding of the different kinds of diffusion and the role of the individual in it. In her discussion of the use of [ʋ] in Leeds, Marsden (2006: 168) states that 31% of the people born between 1960 and 1994 use this incoming variant. She adds that Foulkes and Docherty (2000) find similar results for Derby and New‐ castle where speakers over 45 years of age do not use [ʋ]. As mentioned above, we observe a lag of about a decade for the use of labiodental / r/ between New‐ castle English and Carlisle English for the arrival of the diffusional wave of [ʋ]. However, in the data presented here, there is in fact one female speaker (12F56WC) who was born before 1960 and who uses this feature eight times in the sample. She is a receptionist who is in contact with many young people which might have influenced her choice of variant. The picture for the zero variant in the community is quite blurry. Once again MC female speakers and WC male speakers of the youngest age cohort use this variant to a similar extent. However, the distribution of the zero form for indi‐ viduals is very different from the results for [ʋ]. The distribution is much more homogeneous, i.e. the number of tokens within an age group does not vary dramatically. There is no obvious explanation for this distribution which is not 5.3 (R) 103 triggered by any sociolinguistic factor other than age. This means that the use of Ø is increasing but sex and social class are not predictors for the change. If this is indeed a form diffusing from Newcastle English, the lack of salience of this feature surely plays a role. The lack of significance of sex and social class for the zero form in Carlisle English deviates from the findings in Newcastle English. Foulkes (1997: 84) talks about the Newcastle findings which “vary quite dramatically in correlation with social and stylistic factors.” The increase of the zero form might be an innovation deriving from Newcastle. Foulkes (1997) shows that linking / r/ is almost cate‐ gorically used in Derby but finds high numbers of zero forms for Newcastle speakers. This could indeed be an indicator that Carlisle English is converging with Newcastle English because the zero form diffuses across the far Northern varieties. The overall numbers suggest that the loss of linking / r/ was observed in Newcastle English earlier than in Carlisle English. Earlier data from Carlisle than provided in this study is needed in order to account for this hypothesis. However, these points have to be left for a more comprehensive study of this variant. 5.3.6 Conclusion This chapter has addressed the sociolinguistic distribution of prevocalic / r/ in Carlisle English. The findings have revealed a complex picture for the use of (R) with several changes underway. Moreover, the forms found in this study cor‐ relate with social as well as language-internal factors. A possible influence of Newcastle English on Carlisle English in terms of the incoming variants has been proposed but needs further investigation. Overall, the results presented in this chapter have established a sound base for further investigations into soci‐ olinguistic patterning and geolinguistic distribution of (R). 5.4 (TH) This chapter discusses another fronting process, i.e. TH-fronting. TH-fronting acts as an umbrella term for the merger of / θ/ with / f/ and / ð/ with / v/ . The place of articulation moves from an interdental to a labiodental position in the oral cavity. Similar to the features discussed in §5.2 and §5.3, TH-fronting has been observed in various urban areas in Britain in recent years (cf. Kerswill 2003; Clark and Trousdale 2009; Levon and Fox 2014). 5 Sociolinguistic variables 104 In his overview on recent sound changes in the British Isles, Docherty (2010: 65) comments that “a close second to t-glottaling as the most frequent object of topical comment on the topic of UK English pronunciation is TH-fronting; i.e. the realization of / θ/ and / ð/ as the corresponding labiodentals fricatives [f] and [v].” This chapter discusses to what extent TH-fronting is found in Carlisle Eng‐ lish. In addition, I discuss linguistic and social constraints for the use of these features. 5.4.1 Historical and geographical background The general assumption is that TH-fronting is originally a Cockney feature (Wells 1982: 328). Steele (2008: 212) comments that even though the “collective wisdom” acts from the assumption that TH-fronting originated in the south-east of England, there are some inconsistencies in this argument, e.g. Wells (1982: 328) points out that historically Cockney speakers used stops instead of dental sounds, but he does not specify the timeframe of the change from stops to la‐ biodental fricatives. Nevertheless, based on SED data, Kerswill (2003) assumes that TH-fronting has spread from London, first to Bristol and then to other urban areas in close proximity. He argues that very good train connections between London and Bristol led to an increase of face-to-face contact situations which could have been the reason why TH-fronting was found in Bristol before it was found in other places. On the other hand, TH-fronting is not a feature found in the north of England at the time of the SED and TH-fronting is now diffusing as an in‐ novative form around the country. However, for many varieties information is lacking. For one, we do not know much about the use of TH-fronting in the West Midlands. In addition, hardly any information on the use of TH-fronting is available for Cumbria and large parts of Lancashire or Yorkshire. Yet in recent years a series of sociolinguistic studies have provided us with additional information on the occurrence and diffusion of this feature. Trudgill (1988) comments on the rapid spread of TH-fronting in Norwich. While there is no sign of the use of this feature in the data of his original study in 1968, in the replicated study conducted in 1983, 29% of the speakers already have a full merger, i.e. the voiceless dental fricative is fully replaced by the voiceless labio‐ dental fricative. This “change […] has come completely and dramatically out of the blue” (Trudgill 2002: 56). For the north of England, Wakelin (1972: 98) and Milroy and Milroy (1985: 368) report TH-fronting in Leeds and Sheffield in the 1970s and 1980s respec‐ tively. Trudgill (1994: 33) mentions that “as a feature of younger people’s speech, 5.4 (TH) 105 18 TH-fronting and TH-stopping are also found in other varieties, sometimes in combi‐ nation, i.e. African American Vernacular English, where we find a complex system for (TH), mainly determined by internal constraints, i.e. TH-stopping in word-initial posi‐ tion and TH-fronting in word-final position, e.g. Mufwene et al. (1998); Green (2002). it appears, for instance, to have arrived in Norwich in the 1970s, Sheffield in the 1980s and Exeter in the 1990s. The last we heard it had not yet arrived in New‐ castle or Scotland.” However, Kerswill (2003: 236) dates the arrival of TH-fronting in Newcastle and Glasgow at around 1980. He argues that “the change seems to ‘hit’ very large regions simultaneously, particularly in the far north of England (including Newcastle and Durham) and the central belt of Scotland (represented by Glasgow) at about the same time” (Kerswill 2003: 235). This observation confirms Milroy’s (1996) assumption that the spread of TH-fronting is mainly found in urban areas (Milroy 1996). Three recent studies from the north of England add new information on the spread of TH-fronting. Llamas (2001a) found TH-fronting in Middlesbrough. Her results show that young adult male speakers are leading this change. In contrast to Middlesbrough English, for Darlington English hardly any TH-fronting is reported even though Darlington and Middlesbrough are not even twenty miles apart. Atkinson (2011: 95) reports merely one speaker in Darlington English who uses TH-fronting in only two occasions. Thus, due to lack of variance, (TH) is not considered as a variable in the main study on this north-east variety. TH-fronting only seems to be in its early stages of change in Darlington English. Steele (2008: 223) reports an overall fronting of 5% in her sample in Morley (Leeds). The use of the fronted variant is restricted to the student cohort of her sample. The diffusional change is still in an embryonic stage in Morley (Steele 2008). In Urban Voices (Foulkes and Docherty 1999b) various authors comment on the occurrence of TH-fronting. Watt and Milroy (1999: 30) state that [f] and [v] are found foremost in younger WC speakers in Newcastle but are still used infrequently. Stoddard, Upton and Widdowson (1999: 76) report TH-fronting for Sheffield. This feature is mainly used by males in this variety and for the most part found in word-medial and final position. Mathisen (1999: 111) comments on the occurrence of TH-fronting in Sandwell, a community in the West Mid‐ lands, where adolescents use this incoming variant while adult speakers use the traditional forms [θ] and [ð]. In contrast to the findings in Britain, Hickey (1999: 271f) does not find TH-fronting but TH-stopping 18 in the Irish capital Dublin. On the other hand, Beal (2007: 39) doubts that TH-fronting is really an inno‐ vative feature spreading to the north of England. She states that “there is a need for a more thorough search of nineteenthand early twentieth-century sources 5 Sociolinguistic variables 106 19 Armstrong et al. (1971: 119) provide examples for this sound change, e.g. Thornby is listed as Forneby (1285) and Formeby (1322) and they comment on “the common dialectal confusion between the spirants f and þ.” Ferguson and Robert (1873): 149) provides the example tharf as a variant of tharth. before we conclude that ‘th-fronting’ was absent from Northern dialects until the late twentieth century.” She hypothesises that the number of variation and change studies conducted lately could have led to more attention to this feature. She doubts that TH-fronting is really an innovation in many varieties because she is in particular arguing against Kerswill’s (2003) and Przedlacka’s (2001) hypothesis that TH-fronting is a late 20 th century phenomenon. And indeed, the following quotation about the Cumberland dialect attests that TH-fronting is most likely not a recent innovation in this area. The dental th is not to be found among the ancient words of the district. Kirkby Thore is still pronounced Kirkby Fure, for Thursday we find Furesday, Grisenthwaite is very generally converted into Grislefoot, and smuired appears for smothered (Sullivan 1857: 75). In addition, evidence from typonomy can be provided. Ekwall (1918: 19) men‐ tions that the place name Bewaldeth is recorded as Bowaldif in 1284 which is due to “the common change þ > f [which] is apparently to be assumed.” 19 Thus, TH-fronting could be found (in low frequencies) in Cumberland English. Hence, if TH-fronting occurs in Carlisle English in the present sample, it is not so much an innovation but rather a recycling of this feature (cf. Barnfield and Buchstaller 2010). However, the frequency of this feature must have been very low as neither the SED data for Longtown, nor Ellis’s description of the Edenside dialect gives any instances of TH-fronting. Ellis (1889: 1975), for example, comments that (TH) and (DH) “call for no remark.” Hickey (2002: 1) explains that features can indeed recur during the linguistic history of a variety. For him, transmission, i.e. the acquisition of features during the process of first language acquisition can vary between generations not least due to imperfect learning or dissociation. The transmission of features across generations may be subject to shifts, indeed re‐ versals, which have nothing to do with the relative isolation of the community. These shifts of transmission are referred to collectively in the present paper as ebb and flow. One particular reason for this label is that frequently one has a reversal followed by a reinstatement of a feature value, hence the bidirectionality implied in ebb and flow. 5.4 (TH) 107 Hickey emphasises that unconscious awareness of a feature is an important drive in language change and that linguistic dissociation of these unconscious features from a preceding generation could indeed be the reason why ebb and flow movements are possible. The data analysis in this chapter sheds some more light on a possible ebb and flow process. 5.4.2 Phonetic details The phonetic and phonological details of the two interdental fricatives [θ, ð] are quite complex. Milroy (1996) and Steele (2008) argue that even though [θ] and [ð] are standard forms, they are marked sounds. Thus, a natural change away from an unstable and marked feature could be the language internal source for this change. Maddieson (2011: chapter 19) confirms that interdental sounds are quite uncommon among the world’s languages and therefore are marked. The sound occurs in only 43 (7.6%) of the languages surveyed in the WALS (The World Atlas of Language Structures Online) project. The comparative rarity of sounds of this class suggests that these sounds may also be easily lost over a period of time. This may be because of their quietness, which makes them among the least perceptually salient of consonants (Maddieson 2011, chapter 19). Thus, the low intensity of the interdental sounds in comparison to other sounds could be the source of this change. Another view on the use of [f] instead of [θ] provides the field of First Lan‐ guage Acquisition. The interdental sounds are some of the last to be acquired by young speakers. At the very early stages of acquisition they either replace the sound with a (alveolar/ dental) stop or the labiodental fricative. It is striking that other changes that have been diffusing throughout the country, such as the spread of labiodental / r/ or L-vocalisation, are observed in First Language Ac‐ quisition processes as well. With reference to TH-fronting and L-vocalisation, Britain (2009: 139) states that these forms are “highly typologically unmarked forms […] and acquired early by most Anglophone children, unlike their standard counterparts.” 5.4.3 Data analysis Interviews were recorded as described in chapter 4. Starting at minute ten in the interview, ten tokens for (TH) and ten tokens for (DH) were collected from each speaker. In addition, tokens from the two reading tasks (text and sentence list) 5 Sociolinguistic variables 108 20 TH-fronting is common in all three positions under investigation while DH-fronting is mainly observed in word-internal and final position. were analysed. A total of 1,086 tokens for (TH) and 1,175 tokens for (DH) were collected. The voiced and voiceless realisations have to be analysed separately because of their varying phonotactic behaviour. For (TH) three different environmental factors are considered in the data analysis: initial, medial and final, e.g. that, other, breathe (voiced) and thing, something, north (voiceless). 20 Even though the word-initial voiced context is rarely the subject of TH-fronting, token analysis does include this environment. 5.4.4 Statistical modelling Statistical modelling as described in chapter 5.1 is applied. Factors in the statis‐ tical model are age, sex, social class, speech style (sentence list, text and interview) as external factors and environment (initial, medial and final word position), as well as following sound (consonant, vowel and #) as language internal factor with the individual as random factor and age as continuous factor. 5.4.5 Results In the study five main phonetic variants for (TH) are identified: • the voiceless interdental fricative [θ] and the voiced interdental fricative [ð]; • the voiceless labiodental fricative [f] and the voiced labiodental fricative [v]; • the voiceless dental/ alveolar stop [t]; • the glottal fricative [h]; • the zero form Ø. This includes the deletion of / θ/ , e.g. south [ˈsaʊ]. In the data there are quite a few instances where a voiceless interdental fricative is replaced by a voiced sound and vice versa. The overall results presented in Table 5.4.1 show that the standard variant [θ] is indeed the variant which is used in 90% of all instances while [f] is used 5.1%. More men than women are using [f] and the use of this variant seems to be associated with WC speakers. In terms of age, the variant is mainly used by speakers in the 20-29 age group and the 30-39 age group. 5.4 (TH) 109 [θ] [f] [Ø] [ð] [t] [h] Total Social Class MC N=548 92.3% N=12 2% N=11 1.9% N=20 3.4% N=2 0.3% N=1 0.2% N=594 54.7% WC N=430 87.4% N=43 8.7% N=8 1.6% N=10 2% N=0 0% N=1 0.2% N=492 45.3% Sex M N=483 88.6% N=43 7.9% N=8 1.5% N=10 1.8% N=0 0% N=1 0.2% N=545 50.2% F N=495 91.5% N=12 2.2% N=11 2% N=20 3.7% N=2 0.4% N=1 0.2% N=541 49.8% Age 22-29 N=159 78.7% N=32 15.8% N=5 2.5% N=6 3% N=0 0% N=0 0% N=202 18.6% 30-39 N=106 88.3% N=11 9.2% N=0 0% N=2 1.7% N=0 0% N=1 0.8% N=120 11.1% 40-49 N=138 96.5% N=0 0% N=1 0.7% N=3 2.1% N=1 0.7% N=0 0% N=143 13.2% 50-59 N=201 90.5% N=8 3.6% N=6 2.7% N=7 3.2% N=0 0% N=0 0% N=222 20.4% 60-69 N=298 94.6% N=4 1.3% N=3 1% N=8 2.5% N=1 0.3% N=1 0.3% N=315 29% 70-78 N=76 90.5% N=0 0% N=4 4.8% N=4 4.8% N=0 0% N=0 0% N=84 7.7% Total N=978 90% N=55 5.1% N=19 1.8% N=30 2.8% N=2 0.2% N=2 0.2% N=1,086 100% Table 5.4.1: Absolute numbers and percentages for each variant of (TH) according to social class, sex and age Table 5.4.2 and Table 5.4.3 present the statistical results for the variants [θ] and [f]. The analysis reveals that age, sex and style are all significant social factors, while the following sound represents a linguistic constraint. 5 Sociolinguistic variables 110 Deviance: 633.407 df: 8 Grand mean: 0.895 Speaker ID random standard deviation: 0.834 Application Value: dental frica‐ tive [θ] Factors Log Odds Tokens (N) Propor‐ tion of Ap‐ plication Value Centred Weight Age p<0.01 Continuous 22-78 +1 0.031 Sex p<0.01 F 0.526 544 0.938 0.629 M -0.526 549 0.852 0.371 Style p<0.01 Sentence list 0.714 299 0.940 0.671 Interview -0.214 400 0.892 0.447 Text -0.499 394 0.863 0.378 Following sound p<0.05 Consonant 0.344 214 0.916 0.585 Vowel 0.148 632 0.902 0.537 # -0.492 247 0.858 0.379 Table 5.4.2: Rbrul output for use of voiceless dental fricatives for all speakers Deviance: 279.807 df: 5 Grand mean: 0.05 Speaker ID random standard deviation: 1.326 Application Value: labio‐ dental [f] Factors Log Odds Tokens (N) Propor‐ tion of Ap‐ plication Value Centred Weight Age p<0.01 Continuous 22-78 +1 -0.107 Sex p<0.05 M 0.962 549 0.08 0.723 F -0.962 544 0.02 0.277 5.4 (TH) 111 Social Class p<0.01 WC 1.363 496 0.087 0.796 MC -1.363 597 0.020 0.204 Table 5.4.3: Rbrul output for use of labiodental fricatives for all speakers Table 5.4.4 presents the overall results for (DH), the second variable which is included under the umbrella term TH-fronting. In the study six main phonetic variants for (DH) are identified: • the voiced interdental fricative [ð]; • the voiced alveolar stop [d], which is a traditional variant in Carlisle Eng‐ lish; • the zero form; • the voiceless interdental fricative [θ]; • the voiced labiodental fricative [v] and the voiceless equivalent [f]. [ð] [d] [Ø] [θ] [v] [f] Total Social Class MC N=498 81.9% N=50 8.2% N=52 8.6% N=8 1.3% N=0 0% N=0 0% N=608 51.7% WC N=464 81.8% N=48 8.5% N=33 5.8% N=12 2.1% N=8 1.4% N=2 0.4% N=567 48.3% Gender M N=470 80.9% N=54 9.3% N=41 7.1% N=9 1.5% N=6 1% N=1 0.2% N=581 49.4% F N=492 82.8% N=44 7.4% N=44 7.4% N=11 1.9% N=2 0.3% N=1 0.2% N=594 50.6% Age 22-29 N=169 77.2% N=16 7.3% N=21 9.6% N=3 1.4% N=8 3.7% N=2 0.9% N=219 18.6% 30-39 N=104 83.2% N=14 11.2% N=7 5.6% N=0 0% N=0 0% N=0 0% N=125 10.6% 40-49 N=117 77% N=16 10.5% N=17 11.2% N=2 1.3% N=0 0% N=0 0% N=152 12.9% 50-59 N=208 84.6% N=21 8.5% N=13 5.3% N=4 1.6% N=0 0% N=0 0% N=246 20.9% 60-69 N=286 N=19 5.6% N=24 7.1% N=11 3.2% N=0 0% N=0 0% N=340 28.9% 5 Sociolinguistic variables 112 84.1% 70-78 N=78 83.9% N=12 12.9% N=3 3.2% N=0 0% N=0 0% N=0 0% N=93 7.9% Total N=962 81.9% N=98 8.3% N=85 7.2% N=20 1.7% N=8 0.7% N=2 0.2% N=1,175 100% Table 5.4.4: Absolute numbers and percentages for each variant of (DH) according to social class, sex and age The picture for the (DH) variable is quite complex but the main interest lies in the realisation of / ð/ as [v] or [f] here. Overall, ten tokens of the two variants are found in the sample, all in the group of WC speakers, and all in the age group of 22-29 year-old speakers. Male speakers use the two variants in seven occa‐ sions and females use it in three instances. Due to the very low numbers the results are not run through statistical tests. However, one can derive statements for the two variants by looking at the data. The fronting to [v] is still embryonic in that only very few tokens are found. Individuals in the WC group seem to be the first (early) adopters (cf. Stuart-Smith and Timmins 2010) of this change while males adopt the feature earlier than females in this sample. Only words with medial and word-final position are affected in this sample which is similar to results found in other varieties. If not explicitly mentioned, the results for (DH) are not included in the analysis below. 5.4.5.1 Social factors Age Age is a highly significant factor for the use of [θ] (p<0.01) and [f] (p<0.01) for (TH), i.e. it is most likely that we see a change in progress in apparent time here. The results for age presented in Figure 5.4.1 show the early beginnings of the change. The change from [θ] to [f] can be observed for speakers in their twenties but [f] is already observed in the age group of 30-39 year-old speakers. 5.4 (TH) 113 Figure 5.4.1: Percentage of variants of (TH) according to age Overall twelve tokens of [f] are attested for five speakers who are older than 50 years which was a bit surprising. Yet as was shown before, TH-fronting is not an innovation in Cumbria but low frequencies of this feature are already attested there in the 19 th century due to linguistic instability of interdental fricatives. Hence, these single tokens possibly attest that TH-fronting was indeed already part of the Carlisle English inventory before the diffusional wave with this var‐ iant arrived. It could be that the older speakers adopted the incoming norm. However, there are some indications that this is not the case, i.e. there is no style shift and there is a tendency that older speakers produce [f] in word-final position while younger speakers produce it also in word-initial and word-medial position. Name Age Social class No. of tokens with [f] % 1M22WC 22 WC 20 74.07% 1F22WC 22 WC 4 40.00% 3M31MC 31 MC 7 25.00% 13M59WC 59 WC 5 18.52% 6F27WC 27 WC 4 16.00% 5M36WC 36 WC 3 10.71% 5 Sociolinguistic variables 114 19M68WC 68 WC 3 10.34% 2M26MC 26 MC 3 10.00% 14F57WC 57 WC 2 7.41% 2F23MC 23 MC 1 3.70% 4M34MC 32 MC 1 3.70% 10M52WC 53 WC 1 3.70% 16M62WC 63 WC 1 3.45% Table 5.4.5: Absolute numbers and percentages of [f] for (TH) Table 5.4.5 presents the data for the thirteen speakers who use the labiodental variant on an individual level. Both absolute numbers and percentages for the use of [f] are provided. Four of the speakers who use this feature are females, while nine are male speakers. Four of the speakers have a MC background while the others belong to WC. Two speakers who are over 50 years old are using [f] in over 10% of all instances. In addition, the very strong increase of [f] for the youngest two speakers (1M22WC and 1F22WC) is notable. In general, the use of TH-fronting is increasing in the community but it is still in its early stages. Sex As well as age, sex is a significant factor for the use of [θ] (p<0.01) and [f] (p<0.05). The use of labiodental [f] is a feature preferred more by men than women. Of the overall 55 tokens of [f], only 12 were produced by women while men used 43 tokens with this realisation. Figure 5.4.2 provides the overall dis‐ tribution of variants for male and female speakers. For both sex groups, the use of [f] is still below 10%. 5.4 (TH) 115 Figure 5.4.2: Percentage of variants of (TH) according to sex In other communities it is mainly male speakers who seem to lead this change towards TH-fronting. Llamas (2001a) reports that more male speakers than fe‐ male speakers use the labiodental fricatives, and so do e.g. Williams and Kerswill (1999). Similar results are found in Carlisle: In the very early stages of this change in progress, men are leading the change toward TH-fronting. Social class Unlike other changes discussed here, social class is highly significant (p<0.01) for the use of [f]. Figure 5.4.3 presents the distribution of the four most fre‐ quently used variants for (TH) separated by social class. WC speakers use [f] in 8.7% of all instances while MC speakers use it less often (2%). If we go back to Table 5.4.5, it becomes obvious that the speakers in their fifties and sixties who use [f] are all WC speakers. 5 Sociolinguistic variables 116 Figure 5.4.3: Percentage of variants of (TH) according to social class Once again, this result is similar to results found in other communities in which the WC lead this change in the majority recent case studies. Yet the older speakers who use TH-fronting might not belong to the group who promote the change but their use of [f] might be due to the linguistic predisposition. Style As has been shown for (T) and (R), results obtained by speech style can provide important information about the nature of the change, i.e. whether it is a change from above or below. Table 5.4.6 provides the overall distribution of the data according to speech style. [θ] [f] Ø [ð] [t] [h] Total Style Sen‐ tence list N=281 94.6% N=15 5.1% N=0 0% N=1 0.3% N=0 0% N=0 0% N=297 27.4% Text N=340 87.2% N=19 4.9% N=18 4.6% N=13 3.3% N=0 0% N=0 0% N=390 35.9% Inter‐ view N=357 89.5% N=21 5.3% N=1 0.3% N=16 4% N=2 0.5% N=2 0.5% N=380 36.7% Total 5.4 (TH) 117 N=978 90.1% N=55 5.1% N=19 1.8% N=30 2.8% N=2 0.2% N=2 0.2% N=1,068 100% Table 5.4.6: Absolute numbers and percentages for each variant of (TH) according to style Recall that in total 55 tokens were realised as [f], which is 5.1% of the whole sample. This variant is surprisingly evenly distributed across speech styles with a proportion of around 5% for each style. In fact, speech style is not a statistically significant factor in this change. This result reinforces the assumption that we are observing a change from below. The use of this feature is not consciously monitored in more formal styles by the speakers and therefore one can assume that this change in progress happens unconsciously. 5.4.5.2 Linguistic factors Turning now to the linguistic factor environment within a word (initial, medial and final position), the analysis of this factor can provide us with information about the arrival of TH-fronting in different waves was shown for (T). The overall distribution of (TH) according to position in the word is presented in Table 5.4.7. [θ] [f] Ø [ð] [t] [h] Total Position Initial N=410 91.7% N=20 4.5% N=0 0% N=15 3.4% N=1 0.2% N=1 0.2% N=446 41.2% Medial N=134 86.5% N=4 2.6% N=0 0% N=15 9.7% N=1 0.7% N=1 0.7% N=155 14.3% Final N=434 89.7% N=31 6.4% N=19 3.9% N=0 0% N=0 0% N=0 0% N=481 44.6% Total N=978 90.1% N=55 5.1% N=19 1.8% N=30 2.8% N=2 0.2% N=2 0.2% N=1,086 100% Table 5.4.7: Absolute numbers and percentages for each variant of (TH) according to environment Statistically, the position in the word is not a predictor for the use of [θ] and [f]. However, Figure 5.4.4 makes apparent that the speakers older than 30 primarily 5 Sociolinguistic variables 118 21 Even though 2M26MC, 6F27WC and 3M31MC show an increased use of [f] for (TH), we would not include them in the group of speakers who use the incoming feature. It is more likely that they are a group that resembles a transition stage between the use of a traditional feature and the incoming variant, that is, they have noticed the increased use of [f] by younger speakers and therefore increase their use. However, they have not realised that a change of environment has also taken place. 22 Though probably the use of the traditional feature and the incoming use of [f] overlap and no clear line can be drawn use this variant in final position and only exceptionally in initial position but never in medial position. The three youngest speakers who use the labiodental fricative prefer to use it in initial position rather than in final position. Figure 5.4.4: Use of for (TH) according to individual speaker and environment  21 The distribution of TH-fronting among the speakers in Figure 5.4.4 provides us with additional information about the change: Two separate processes of TH-fronting seem to be observable. 22 On the one hand, the use of TH-fronting in word-final position which is mainly used by older speakers and on the other hand the use of [f] in initial position in particular by the two youngest speakers. This result might be interpreted as two different kinds of TH-fronting. 5.4.6 Discussion Overall, the interdental forms [θ] and [ð] are still very much the norm in Carlisle English. The speech of 13 out of the 40 speakers analysed here have instances 5.4 (TH) 119 of TH-fronting but none of the speakers categorically uses the labiodental form [f]. The speaker who uses the labiodental form most often has a rate of about 80%. Thus, the rate is far lower than in communities such as Derby in the 1990s (cf. Foulkes and Docherty 1999a) but interestingly, the rate of TH-fronting seems to be higher in Carlisle English than in Darlington English (Atkinson 2011). A crucial finding here is that not only do young speakers use this variant, but five of these speakers are actually older than 50. As was mentioned before, TH-fronting in low frequency is traditionally attested in Cumbria. Thus, it might be that the older speakers are using TH-fronting but the motivation for the use of this variant might be different for older than for younger speaker. This view is enhanced by the fact that the older speakers mainly use it in word-final po‐ sition rather than in initial or medial position. In comparison, younger speakers use the labiodental variant in word-initial position in addition to the other two positions. TH-fronting is quite salient in word-initial position (and medial position). The use of [f] instead of [θ] in word-final position seems to be less salient, which is why the use of TH-fronting could go mostly unnoticed. However, even though initially the use of TH-fronting by older speakers was thought to be caused by linguistic factors, TH-fronting seems to be indexical for WC speakers in general. In low frequencies this feature was already found in Carlisle English but it seems that a wave of diffusion arrived in the community seen by the use of TH-fronting by the youngest speakers. Given that older speakers use TH-fronting but do not seem to be affected by the same processes as the younger speakers, we cannot assume that it is an innovation in this urban dialect. It is more of an ebb and flow movement (Hickey 2002). Younger speakers simply ‘recycle’ a feature which has been found in the community before. The increase of the use of [f] for younger speakers is most likely due to the spread of this variant across the country. The use of labiodental fricatives for interdentals expands its scope in terms of voicing and environ‐ ments which confirms the assumption that the use of this feature has a different motivation than the TH-fronting which we observe for older speakers. Dating the arrival of TH-fronting in Carlisle English as diffusional feature, it is likely that this form arrived in Carlisle English around the late 1990s/ beginning of the 2000s. If our assumption is right, the acquisition of the new form of TH-fronting took place shortly before 1F22WC and 1M22WC were adolescents, 1995-2000 would be a realistic guess for the arrival of this feature in the community. This also means that the feature arrived in Carlisle about fifteen years later than it was attested in Newcastle (cf. Kerswill 2003). 5 Sociolinguistic variables 120 In contrast to the results for (TH), the use of [v] for (DH) indeed seems to be an (embryonic) innovation to the community. None of the older speakers use this variant and only the two youngest speakers show variation between [ð] and [v] Hence, due to the diffusional wave both interdental sounds are affected by the change and the results imply that TH-fronting and DH-fronting did not happen parallel. The results provided in this chapter clearly show that TH-fronting has started to increase after a period of sporadic use in Carlisle English. Evidently, not all of the younger speakers are using this variant; it is associated in particular with WC. The fact that hardly any variation between styles is observable leads to the conclusion that this must be a change from below, i.e. below the level of con‐ sciousness. 5.4.7 Conclusion TH-fronting as a change from below is attested for Carlisle English. However, unlike other studies, this study argues that the use of [f] for (TH) is not a com‐ plete innovation in this community. Nevertheless, it has to be acknowledged that the diffusional wave reached Carlisle in the early 2000s. The heaped use of [f] in initial and medial position, instead of final position which seems to be more traditional, testifies this. 5.4 (TH) 121 1 As pointed out earlier, the discussion whether G O O S E -fronting is a diffusional feature is avoided so far in research but the results from other urban areas do suggest that it is indeed a feature that is spreading across the country. 6 Discussion In the following, the findings from the previous chapters are brought together and set into context in order to gain an understanding of levelling and diffusion processes in Carlisle English. Variation and change processes in this variety are discussed based on the research questions defined at the outset of this thesis. Which geolinguistic processes are observable? Levelling as well as diffusion of linguistic features can be observed in Carlisle English. The former is observed in the levelling of prevocalic / r/ , where the local tap variant is replaced by the supraregional approximant realisation. Diffusion of T-glottaling, labiodental / r/ and TH-fronting are also attested in Carlisle Eng‐ lish. Table 6.1 summarises the results from §5 and provides an overview of diffu‐ sional features for each individual in the sample. 1 Given the distribution within the community, the very first of the consonantal variants to have arrived in this urban dialect was T-glottaling. All speakers have some degree of T-glottaling, even though it is not attested for everybody in all environments studied in §5.2. The data implies that WC speakers started using the intervocalic variant first but then the feature was reinterpreted and used by both MC and WC speakers, yet none of them use the variant categorically. Intervocalic position is the last environment where T-glottaling was introduced. The use of labiodental / r/ must have arrived in Carlisle English a little later than intervocalic T-glottaling. Again, none of the speakers in the sample use this feature categorically. The last variant that is discussed in §5 is TH-fronting. As argued in §5.4, the speakers who use it in word-final position display a tradi‐ tional form while the two youngest speakers use TH-fronting in all positions and therefore use the incoming form. None of the speakers uses TH-fronting categorically. Hence, the three consonantal variants that are spreading across England/ Britain have been attested in Carlisle English now as well but none of them is yet used categorically. In apparent time, T-glottaling is the feature that has been attested the longest in the speech community while labiodental / r/ is frequently 2 For T-glottaling, a direct comparison is not possible. used by speakers younger than 45 years old and TH-fronting is mainly used by the youngest speakers of the sample. At least labiodental / r/ and TH-fronting arrived in Carlisle later than in Newcastle, 2 which is unsurprising taken that researchers assume that these features spread in waves or they spread to bigger cities first before they arrive in smaller towns (cascade diffusion). However, in order to find out which model applies here, data from places between Newcastle and Carlisle would be needed. If smaller towns and villages between these two cities are more advanced in their use of these features, one can assume that the diffusion spread in waves. If this is not the case and Carlisle speakers are more advanced in their use of these consonantal features, the cascade model applies. Speaker G O O S E -fronting Intervocalic T-glottaling Labiodental / r/ TH-fronting 1F22WC ✓ ✓ ( ✓ ) ✓ 1M22WC ✓ ✓ ✓ ✓ 2F23MC ✓ ✓ ✓ ( ✓ ) 3F24MC ✓ / ✓ / 4F25MC ✓ ✓ ✓ / 5F25MC ✓ ✓ ✓ / 2M26MC ✓ / ✓ ✓ 6F27WC ✓ ✓ ✓ ✓ 3M31MC ✓ ( ✓ ) ✓ ✓ 4M34MC ✓ ✓ ✓ ( ✓ ) 7F36MC ✓ ( ✓ ) / / 5M36WC ✓ ✓ ( ✓ ) ✓ 6M38WC ✓ ✓ ✓ / 7M41MC ✓ ( ✓ ) / / 8F42MC ✓ ( ✓ ) ✓ / 8M43WC ✓ / / / 9F47MC ✓ / / / 6 Discussion 123 10F49WC ✓ / / / 9M52WC ✓ ✓ / ( ✓ ) 11F53MC ✓ / / / 10M55WC ✓ ✓ ( ✓ ) / 12F56WC ✓ / ✓ / 13F57WC ✓ / ( ✓ ) ( ✓ ) 11M57MC ✓ / / / 14F59WC ✓ / ( ✓ ) / 12M59WC ✓ / ( ✓ ) ✓ 15F60MC ✓ / / / 13M60MC / / / / 16F61WC ✓ ✓ / / 14M61MC ✓ / ( ✓ ) / 15M62WC ✓ / / ( ✓ ) 17F63WC ✓ / / / 16M63WC / / / / 17M67MC ✓ / / / 18F67MC ✓ / / / 19F68MC ✓ / ( ✓ ) / 18M68WC / / / ✓ 20F71WC / / / / 19M73MC ✓ / / / 20M78MC ✓ / / / Table 6.1: Overview of diffusional features found in Carlisle English (check marks in brackets mean that the feature was used twice at the most) Thus, the comparatively late attestation of the consonantal features in apparent time discussed here for Carlisle English confirms the hypothesis that the features spread across the country first and foremost by diffusion due to face-to-face contact. 6 Discussion 124 A third process which was observed is resistance to change. The results in §5.1 suggest that variation exists for the realisation of the G OAT vowel; the var‐ iation is however stable and does not change in apparent time. As briefly men‐ tioned before, the back G OAT vowel seems to be an identity marker of the accent and the non-participation in the supraregional process of G OAT -fronting leaves us with the assumption that this traditional feature is retained. In addition, G OAT -fronting as we see in the south of England cannot occur when the vowel is realised as a monophthong, i.e. internal factors inhibit the fronting of this vowel (cf. Jansen 2017a). It is important to stress that not all features that are found in Carlisle English due to diffusion are innovations as Kerswill (2003) claims for the north of Eng‐ land. In §5.4 I provide evidence that fronting of (TH) was already part of the Carlisle English inventory before the diffusional waves of (TH) and (DH) arrived. Beal’s (2007) claim for caution and her prompt to include adult speakers in a sample of a variation and change study, as done here, is an important part of this argument. How does linguistic variation pattern for the social categories age, sex and social class in CE? The social variables analysed in §5.1-5.4 are discussed in the light of age, sex and social class. Age is the most important factor in language change. Indeed, age is a predictor for change in apparent time for the G O O S E variable as well as for the incoming variants of (T), (R) and (TH). Age also plays a role for the variants that are decreasing, e.g. [t] for (T) and [ɾ] for (R). Hence, these are variants where change in progress can be observed. However, while variation does exist for G OAT , age is not a significant factor for change and thus, this variable is not undergoing change at the moment. In general, age is the decisive factor to observe language change but in the case of (TH), age as a social factor provides us with even more information. Beal (2007) criticises the fact that most variation and change studies that deal with features that are diffusing around the country focus on adolescents but do not take adult speakers into account. She argues that it is important to include this group of speakers in order to see whether any of the diffusing features was already present in the variety before the latest wave of diffusion arrived. This is indeed the case in Carlisle English. The data shows that there is a sporadic use of TH-fronting for adults older than thirty, mainly in word-final position. This finding from the present study in combination with the historical evidence of TH-fronting in Cumberland (Sullivan 1857, Ferguson 1873) and the evidence from place name research provides us with a convincing argument that TH-fronting is not an innovation in Carlisle English. Nevertheless, the data from 6 Discussion 125 the sociolinguistic interviews also shows that a new wave of TH-fronting has arrived in the community. The main argument for this being an incoming wave rather than a change which happened within the community is the fact that the position of fronting within the word changes, i.e. the incoming TH-fronting, is found in word-initial, word-medial and less in word-final position while older speakers mainly use it in word-final position and in a few cases in word-medial position. In addition, DH-fronting is not found in older speakers but the younger speakers who use TH-fronting in all positions also use DH-fronting. This can be seen as the second argument for a new incoming wave of this feature. As mentioned before, Hickey’s (2002) model of ebb and flow and the hypothesis that features are ‘recycled’ stem from a similar idea: features can rise to become social variables even though they have been around for some time in a variety in low numbers, e.g. Barnsfield and Buchstaller (2010) find that the intensifier dead has been present in Tyneside English in single instances but only recently has it increased in Newcastle English. This observation is quite similar to the observation made here for TH-fronting. In the ebb and flow model, Hickey re‐ traces the life cycle of a feature. The frequency of a feature is not constantly high but can vary. If the feature does not carry much social meaning it can still survive in a variety in low quantities. It is likely that this feature then increases due to a trigger from outside as was seen in the data. Here, the diffusional wave of TH-fronting served as an external trigger for the increase of this feature. However, the use of TH-fronting primarily by male and WC speakers does imply that this traditional variant had not completely lost its social meaning. Table 6.2 provides an overview of groups which are leading the different changes. It shows that women and MC speakers are promoting the change to‐ wards G O O S E -fronting. The same two groups are leading the change towards the use of labiodental / r/ . Females also are leading the change towards the use of glottal stops. Striking is that young MC male speakers do not take part in the change towards the use of intervocalic glottal stops. G O O S E -fronting [ʔ] (general) [ʔ] (intervo‐ calic) [ʋ] [f] F ✓ ✓ ✓ ✓ / M / / / / ✓ MC ✓ / ✓ ✓ / 6 Discussion 126 WC / / ✓ / ✓ Table 6.2: Groups who are leading a change Table 6.2 clearly shows that women are the motor of change for variants that carry overt prestige, e.g. for G O O S E -fronting (§5.1), for the glottal stop variant of (T) (§5.2) and the labiodental variant of (R) (§5.3). From the data in §5.4 for TH-fronting one cannot deduce that sex is a factor for the incoming change yet but there are indications that this is most likely a change from below. Male speakers on the other hand tend to retain traditional or vernacular fea‐ tures. In particular the retention of the conservative tap variant for (R) by WC male speakers in §5.3 confirms this observation also made by others (e.g. Trudgill 1974). In fact, for the tap variant one sees an interplay of sex and social class as external factors. For a long time, social class has been said to be the strongest factor in language variation studies (e.g. Labov 1972b: 212). However, Watt (2000) claims that in Newcastle English sex has become a much stronger factor than social class. This seems to be only partially true for Carlisle English. Social class cannot be claimed to be the strongest factor for social stratification. Instead, sex and social class have an almost equal share in the social stratification of the incoming features discussed here. It was assumed that the specific conditions in the city of Carlisle might have an influence on the distribution of social class within the community. Foremost, Carlisle is a working-class town with a high percentage of working-class people. The social mixing between WC and MC speakers is quite high. There are only very few areas in Carlisle which would be identified as mainly WC or MC area. Thus, the social classes are not isolated from each other as is the case in the Belfast study (Milroy 1987) and the Wilmslow study (Watts 2005). I argue that dense networks which are said to be quite typical of WC communities are less dense in Carlisle. A further factor which breaks up dense networks is people moving between different estates. Those who filled in the questionnaire had lived in two or more different areas of the city. In addition, The Lanes - one of the oldest, most densely populated settlement areas of Carlisle - and some other estates were demolished in the 1980s and 1990s. As a result, the social networks that existed in this area broke open which can allow for innovations to spread faster. However, the analysis chapters of the sociolinguistic variables have shown that variation is in fact stratified according to social class. Therefore, even though the social class division is not very sharp in Carlisle, linguistically, 6 Discussion 127 one observes stratification according to social class, even though it is not the most decisive social factor. Is variation leading to language change? If so, are these changes led by internal or external motivation? The analysis of G O O S E , (T), (R) and (TH) indeed has shown that variation leads to change in these variables. Due to the diffusional nature of these changes around the country, one can state with certainty that the changes towards in‐ coming variants is triggered by external motivation. In addition, for all four variables a change in progress is attested accompanied by an increase of varia‐ bility. However, the analyses have also shown that the changes are always ac‐ companied by complex interplays of internal and external factors. For F OOT no change is observed or expected, yet for G OAT the F2 value varies which has not led to change yet. The traditional vowel which is realised quite far back is retained. Unlike other varieties, a fronting has not set in. The resist‐ ance to change could most likely has language internal reasons as the fronting of goat seems to be blocked by the nature of the vowel, i.e. varieties with mono‐ phthongal G OAT vowels tend not to show fronting tendencies (cf. Jansen 2017a). Conclusion In this chapter I have drawn together the findings of the analysis chapters and have returned to the original research questions of this thesis. In considering a wide range of external but also internal variables, I have uncovered a complex picture of the different types of variation and change processes that are present in this modern urban British speech community. 6 Discussion 128 7 Conclusion This thesis has investigated language variation and change processes in Carlisle English by applying auditory and acoustic means for analysing variation. I con‐ ducted a sociolinguistic investigation of the variety of English spoken in Carlisle, in order to establish the linguistic mechanisms and outcomes of social and geo‐ graphical change in a peripheral urban English community. I specifically fo‐ cussed on the mechanisms of diffusion of externally originating features as well as levelling. By taking into account various internal as well as external factors, I have shown that both geolinguistic processes - levelling and diffusion - are at work in Carlisle English. The study set out to determine whether the features that have been diffusing around the country for some time now are also attested in Carlisle English. This is of particular interest because Carlisle is the city in England that is furthest away from London, the place where the consonantal features discussed here are assumed to have originated. The results provide further evidence for the diffu‐ sion of features around the country and their delayed occurrence in Carlisle English. The use of the consonantal features discussed in §5.2-§5.4, in particular the use of TH-fronting, is still embryonic and none of the speakers use this feature categorically. In addition, I argue that not only consonants are diffusing but also vowels, in particular the G O O S E vowel. The book contributes to the ongoing research into variation and change in British English and in particular Northern English, by illustrating the patterns of heterogeneity observable in this Cumbrian city variety. Our understanding of contact-induced change in British urban varieties has largely overlooked English in Cumbria. The analysis of Carlisle English contributes to our growing understanding of the patterns of language variation and change observable across the UK. Areas of future research Findings in this study have shown that a more careful approach to ‘innovations’ in a variety is needed. As the analysis of TH-fronting has revealed and as Beal (2007) has shown for other features in the north of England, it is worth digging for historical dialect descriptions in order to find out which features are actually innovations and which are not. In addition, there is potential for future research by investigating morphosyntactic and discourse features in a variation and change framework. In particular the latter have not been investigated in the north-west of England. Further work is also required to establish a more detailed account of the lin‐ guistic relationship of Carlisle English and Tyneside English. Newcastle is the centre in the north-east and the most important place for people in Carlisle to go to and thus Carlisle English speakers are frequently in contact with Newcastle English speakers. It is very likely that this regular dialect contact has repercus‐ sions on the Carlisle English dialect. However, as was shown, the G O O S E -fronting observed in Carlisle English cannot have diffused from Tyneside English, be‐ cause this change was not observable for speakers in the north-east at the time when the Carlisle speakers were recorded. Thus, a structured comparison of the two varieties is desirable which also includes perception and attitude tests. A real-time study of Carlisle English would provide a more complex under‐ standing of the linguistic developments in the variety, in particular the levelling processes in the late 19 th and 20 th century. Finally, a county-wide study of the English of Cumbria would provide us with more information on the centre/ periphery dichotomy which is often employed in sociolinguistics. The relationship between Carlisle and the communities on the west coast, e.g. Workington, Whitehaven and Maryport is of particular in‐ terest. These are fairly isolated areas which could provide further information about diffusion processes. Jansen (2018) is a first approach to investigating lan‐ guage variation and change in Maryport as peripheral community. The possible influence of Carlisle English on these varieties can provide more theoretical implications for understanding geolinguistic processes such as diffusion and levelling. Attitudinal studies would have to be an imminent part of such a study due to complex identity and social practice structures. Hence, Carlisle English and Cumbrian English still offer a wide range of possible research projects for the future. 7 Conclusion 130 References Altendorf, Ulrike and Dominic Watt. 2008. The dialects in the south of England. In Bernd Kortmann and Clive Upton (eds.) Handbook of varieties: The British Isles. Berlin: Mouton de Gruyter. 178-203. Armstrong, A. M., A. Mawer, F. M. Stenton and B. Dickins. 1971. The place-names of Cumberland. Cambridge: Cambridge University Press. Armstrong, Anita. 2003. Were the Irish in the north-west a 'race apart? Aspects of Irish migration into the north-west of England in the nineteenth century concentrating on the half century 1820-1870 with a micro case study of Irish settlement in the Caldewgate area Carlisle. MA module. Lancaster: University of Lancaster. Atkinson, John. 2011. Linguistic variation and change in a north-east border town: a soci‐ olinguistic study of Darlington. PhD thesis. Newcastle: Newcastle University. Auer, Peter. 2007. Mobility, contact and accommodation. In Carmen Llamas, Louise Mul‐ lany and Peter Stockwell (eds.) The Routledge companion to sociolinguistics. London: Routledge. 109-115. Auer, Peter, Birgit Barden and Beate Grosskopf. 1998. Subjective and objective parameters determining 'salience' in long-term dialect accommodation. Journal of Sociolinguistics 2: 163-187. Bailey, Guy, Tom Wikle, Jan Tillery and Lori Sand. 1993. Some patterns of linguistic diffusion. Language Variation and Change 5: 359-390. Baranowski, Maciej. 2007. Phonological variation and change in the dialect of Charleston, South Carolina. Durham, N.C.: Duke University Press. Baranowski, Maciej. 2008. The fronting of the back upgliding vowels in Charleston, South Carolina. Language Variation and Change 20: 527-551. Baranowski, Maciej. 2017. Class matters: The sociolinguistics of G O O S E and G O AT in Man‐ chester English. Language Variation and Change 29: 301-339. Barnfield, Kate and Isabelle Buchstaller. 2010. Intensifiers in Tyneside: Longitudinal de‐ velopments and new trends. English World-Wide 31: 252-287. Beal, Joan C. 2007. ‘To explain the present’: 18th and 19th-century antecedents of 21st-century levelling and diffusion. In Javier Pérez-Guerra, Dolores González-Ál‐ varez, Jorge Bueno-Alonso and Esperanza Rama-Martínez (eds.) “Of varying language and opposing creed”: New insights into Late Modern English. Bern: Lang. 25-46. Beal, Joan C. 2008. English dialects in the North of England: phonology. In Bernd Kort‐ mann and Clive Upton (eds.) Handbook of Varieties: The British Isles. Berlin: Mouton de Gruyter. 122-144. Beal, Joan C. 2010. An introduction to regional Englishes: Dialect variation in England. Edinburgh: Edinburgh University Press. Beal, Joan C., Lourdes Burbano-Elizondo and Carmen Llamas. 2012. Urban North-Eastern English: Tyneside to Teesside. Edinburgh: Edinburgh University Press. Beal, Joan C. and Karen P. Corrigan. 2002. Relativisation in Tyneside and Northumbrian English. In Patricia Poussa (ed.) Relativisation on the North Sea littoral. München: Lincom Europa. 125-134. Beal, Joan C. and Karen P. Corrigan. 2009. The impact of nineteenth-century Irish English migrations on contemporary northern Englishes: Tyneside and Sheffield compared. In Esa Penttilä and Heli Paulasto (eds.) Language contacts meet English dialects: Studies in honour of Markku Filppula. Newcastle upon Tyne: Cambridge Scholars. 231-258. Bell, Allan. 1984. Language style as audience design. Language in Society 13: 145-204. Boberg, Charles. 2000. Geolinguistic diffusion and the US-Canada border. Language Var‐ iation and Change 12: 1-24. Boersma, Paul and David Weenink. 2008. Praat: Doing phonetics by computer. http: / / www.fon.hum.uva.nl/ praat/ Boughton, Zoë. 2005. Accent levelling and accent localisation in northern French: Com‐ paring Nancy and Rennes. Journal of French Language Studies 15: 235-256. Brilioth, B. A. 1913. A grammar of the dialect of Lorton (Cumberland). London: Oxford University Press. Britain, David. 2002. Diffusion, levelling, simplification and reallocation in past tense BE in the English Fens. Journal of Sociolinguistics 6: 16-43. Britain, David. 2005. The dying dialects of England? In Antonio Bertacca (ed.) Historical linguistic studies of spoken English: Papers read at the 11th Italian Conference on the History of the English Language, Pisa, 5-7 June 2003. Pisa: Edizioni Plus. 35-46. Britain, David. 2009. One foot in the grave? Dialect death, dialect contact, and dialect birth in England. International Journal of the Sociology of Language 196-197: 121-155. Britain, David. 2010a. Language and space: The variationist approach. In Peter Auer and Jürgen E. Schmidt (eds.) Language and space: An international handbook of linguistic variation. Volume 1: Theories and methods. Berlin: Mouton De Gruyter. 142-163. Britain, David. 2010b. Supralocal regional dialect levelling. In Carmen Llamas and Do‐ minic Watt (eds.) Language and identities. Edinburgh: Edinburgh University Press. 193-204. Broadbent, Judith M. 2008. t-to-r in West Yorkshire English. English Language and Lin‐ guistics 12: 141-168. Buchstaller, Isabelle. 2015. Exploring linguistic malleability across the life span: Age-spe‐ cific patterns in quotative use. Language in Society 44: 457-496. References 132 Burbano-Elizondo, Lourdes. 2006. Regional variation and identity in Sunderland. In Tope Omoniyi and Goodith White (eds.) The sociolinguistics of identity. London: Continuum. 113-128. Burbano-Elizondo, Loudes. 2012. Sociolinguistic variation and change in F A C E and G O AT in Sunderland. Paper presented at the Northern Englishes Workshop 5. Nottingham: Nottingham Trent University. Carlisle City Council Equality Impact Assessment. Available online: http: / / www.carlisle-city.gov.uk/ pdf/ CP%20Econ%20Strat%20Impact%20Assessment.pdf (ac‐ cessed: 22 December 2011). Census 2001: Key Statistics Carlisle. Available online: http: / / www.neighbourhood.statis‐ tics. gov.uk/ dissemination (accessed: 22 December 2011) Chambers, J. K. and Peter Trudgill. 1998. Dialectology, 2 nd edition. Cambridge: Cambridge University Press. Chambers, J. K. 2009. Sociolinguistic theory: Linguistic variation and its social signifi‐ cance. Malden, MA: Wiley-Blackwell. Clark, Lynn and Graeme Trousdale. 2009. Exploring the role of token frequency in pho‐ nological change: Evidence from TH-Fronting in east-central Scotland. English Lan‐ guage and Linguistics 13: 33-55. Coggle, Paul. 1993. Do you speak Estuary? London: Bloomsbury. Coombes, M. 1995. The impact of international boundaries on labour market area defi‐ nitions. Area 27: 46-52. Cooper, Paul. 2017. 'Turtlely amazing': the enregisterment of 'Yorkshire' dialect and the possibility of G O A T fronting as a newly-enregistered feature. In Chris Montgomery and Emma Moore (eds.) Language and a sense of place: Studies in language and re‐ gion. Cambridge: Cambridge University Press. 348-367. Coupland, Nikolas. 1984. Accommodation at work: Some phonological data and their implications. International Journal of the Sociology of Language 46: 49-70. Cruttenden, Allan. 2008. Gimson's pronunciation of English, 7 th edition. London: Hodder Education. Deterding, David. 1997. The formants of monophthong vowels in Standard Southern British English pronunciation. Journal of the International Phonetic Association 27: 47- 55. Di Paolo, Marianna and Malcah Yaeger-Dror. 2010. Introduction. In Marianna Di Paolo and Melcah Yaeger-Dror (eds.) Sociophonetics: A student's guide. London: Routledge. 1-6. Docherty, Gerard. 2010. Phonological innovation in contemporary spoken British Eng‐ lish. In Andy Kirkpatrick (ed.) The Routledge Handbook of World Englishes. London: Routledge. 59-75. References 133 Docherty, Gerard and Paul Foulkes. 1999. Derby and Newcastle: Instrumental phonetics and variationist studies. In Paul Foulkes and Gerard Docherty (eds.) Urban voices: Accent studies in the British Isles. London: Edward Arnold. 47-71. Docherty, Gerard and Paul Foulkes. 2005. Glottal variants of / t/ in the Tyneside variety of English. In William J. Hardcastle and Janet Beck (eds.) A Figure of Speech: A Fes‐ tschrift for John Laver. Mahwah: Lawrence Erlbaum Associates. 173-199. Drummond, Rob. 2011. Glottal variation in / t/ in non-native English speech: Patterns of acquisition. English World-Wide 32: 280-308. d-maps.com. Cumbria. Available online: http: / / d-maps.com/ carte.php? num_car=92551&lang=de d-maps.com. United Kingdom. Available online: http: / / d-maps.com/ carte.php? num_car=14447&lang=de Ekwall, Eilert. 1918. Scandinavians and Celts in the North-West of England. Lund: Gleerup. Ellis, Alexander J. 1889. On early English pronunciation. Available online: https: / / archive.org/ details/ onearlyenglishpr00elliuoft Fabricius, Anne. 2002. Ongoing change in modern RP: Evidence for the disappearing stigma of t-glottalling. English World-Wide 23: 115-136. Fabricius, Anne. 2007. Vowel formants and angle measurements in diachronic sociopho‐ netic studies: F O O T -fronting in RP. In Jürgen Trouvain and William Barry (eds.) Pro‐ ceedings of the 16th International Congress of Phonetic Sciences. Saarbrücken: Univer‐ sität des Saarlandes. 1477-1480. Fabricius, Anne, Dominic Watt and Daniel Ezra Johnson. 2009. A comparison of three speaker-intrinsic vowel formant frequency normalization algorithms for sociopho‐ netics. Language Variation and Change 21: 413-435. Feagin, Crawford. 2004. Entering the community: Fieldwork. In J. K. Chambers, Peter Trudgill and Natalie Schilling-Estes (eds.) The Handbook of language variation and change. Oxford: Blackwell. 20-39. Ferguson, Robert. 1873. The dialect of Cumberland: with a chapter on its place-names. London: Williams and Norgate. Ferragne, Emmanuel and François Pellegrino. 2010. Formant frequencies of vowels in 13 accents of the British Isles. Journal of the International Phonetic Association 40: 1-34. Finnegan, Katie. 2009. G O A T -fronting in Sheffield English. Poster presented at UK Lan‐ guage Variation and Change 7. Newcastle: Newcastle University. Flemming, Edward. 2003. The relationship between coronal place and vowel backness. Phonology 20: 335-373. Flynn, Nicholas. 2011. Comparing vowel formant normalisation procedures. York Papers in Linguistics 11: 1-28. Flynn, Nicholas. 2012. A sociophonetic study of Nottingham speakers. PhD thesis. York: University of York. References 134 Forsyth, Harold. 2002. Hoo's ta gaan on? A book of Cumberland Tales, in the dialect of Carlisle and the district, by the late Harold Forsyth. Lakeland Dialect Society. Foulkes, Paul. 1997. English [r]-sandhi a sociolinguistic perspective. Histoire, Epistémo‐ logie, Langage: 73-96. Foulkes, Paul and Gerard Docherty. 1999a. Urban voices an overview. In Paul Foulkes and Gerard Docherty (eds.) Urban voices: Accent studies in the British Isles. London: Edward Arnold. 1-24. Foulkes, Paul and Gerard Docherty (eds.) 1999b. Urban voices: Accent studies in the British Isles. London: Edward Arnold. Foulkes, Paul and Gerard Docherty. 2000. Another chapter in the story of / r/ : ‘Labiodental’ variants in British English. Journal of Sociolinguistics 4: 30-59. Foulkes, Paul and Gerard Docherty. 2007. Phonological variation in England. In David Britain (ed.) Language in the British Isles. Cambridge: Cambridge University Press. 52- 74. Foulkes, Paul, Gerard Docherty and Mark Jones. 2010. Analyzing stops. In Marianna Di Paolo and Malcah Yaeger-Dror (eds.) Sociophonetics: A student's guide. London: Rout‐ ledge. 58-71. Giegerich, Heinz J. 1992. English phonology: An introduction. Cambridge: Cambridge Uni‐ versity Press. Giles, Howard and Peter F. Powesland. 1975. Speech style and social evaluation. London, New York: Academic Press. Green, Lisa. 2002. African American English: A linguistic introduction. Cambridge: Cam‐ bridge University Press. Haddican, Bill, Paul Foulkes, Vincent Hughes and Hazel Richards. 2013. Interaction of social and linguistic constraints on two vowel changes in northern England. Language Variation and Change 25: 371-403. Hall, Damien. 2008. A sociolinguistic study of the regional French of Normandy. PhD thesis. Philadelphia: University of Pennsylvania. Hall-Lew, Lauren. 2009. Ethnicity and phonetic variation in a San Francisco neighbor‐ hood. PhD thesis. Stanford: Stanford University. Harrington, Jonathan, Sallyanne Palethorpe and Catherine I. Watson. 2000. Does the Queen speak the Queen's English? Nature 408: 927-928. Henton, Caroline G. 1983. Changes in the vowels of Received Pronunciation. Journal of Phonetics 11: 353-371. Hickey, Raymond. 1999. Dublin English: current changes and their motivation. In Paul Foulkes and Gerard Docherty (eds.) Urban voices: Accent studies in the British Isles. London: Edward Arnold. 265-281. Hickey, Raymond. 2000. Dissociation as a form of language change. European Journal of English Studies 4: 303-315. References 135 Hickey, Raymond. 2002. Ebb and flow: A cautionary tale of language change. In Teresa Fanego, Belén Mendez-Naya and Elena Seoane (eds.) Sounds, words, texts, change: Selected papers from the Eleventh International Conference on English Historical Lin‐ guistics (11 ICEHL). Amsterdam: John Benjamins. 105-128. Hickey, Raymond. 2004. Development and Diffusion of Irish English. In Raymond Hickey (ed.) Legacies of colonial English. Cambridge: Cambridge University Press. 82-117. Hickey, Raymond. 2005. Dublin English. Amsterdam: John Benjamins. Hickey, Raymond. 2007. Irish English: History and present-day forms. Cambridge: Cam‐ bridge University Press. Hickey, Raymond. 2012a. Internally and externally motivated language change. In Juan M. Hernández-Campoy and Juan C. Conde-Silvestre (eds.) The Handbook of historical sociolinguistics. Chichester: Wiley. 387-407. Hickey, Raymond. 2012b. Supragregionalisation. In Alexander Bergs and Laurel J. Brinton (eds.) English historical linguistics: An international handbook. Berlin: de Gruyter. 2060- 2075. Hinskens, Frans. 1996. Dialect levelling in Limburg. Structural and sociolinguistic aspects. Tübingen: Max Niemeyer. Hughes, Arthur, Peter Trudgill and Dominic Watt. 2012. English accents and dialects. London: Hodder Arnold. Huntington, Fiona and Caoimhe McKerr. 2008. Carlisle and district health improvement and health inequalities strategy 2008 - 2010. Baseline Assessment and City Profile. Available online: http: / / www.carlisle.gov.uk/ pdf/ CP%20Draft%20City%20Profile%201. Jansen, Sandra. 2013. ‘I don’t sound like a Geordie! ’: Phonological and morphosyntactic features of Carlisle English. In Nils-Lennart Johannesson, Gunnel Melchers and Beyza Björkmann (eds.) Of butterflies and birds, of dialects and genres: Essays in honour of Philip Shaw. Stockholm: Acta Universitatis Stockholmiensis. 209-224. Jansen, Sandra. 2014. Salience effects in the north-west of England. Linguistik Online 66: 91-110. Jansen, Sandra. 2015a. A century of change in (r) in Carlisle English. Internal constraints in a levelling process. In Eivind Torgersen, Stian Hårstad and Brit Mæhlum (eds.) Language Variation - European Perspectives V: Selected papers from the Seventh Inter‐ national Conference on Language Variation in Europe (ICLaVE 7), Trondheim, June 2013 (Studies in language variation). Amsterdam: John Benjamins. 129-143. Jansen, Sandra. 2015b. Carlisle and Cumbria. In Raymond Hickey (ed.) Researching northern English. Amsterdam: John Benjamins. 205-226. Jansen, Sandra. 2017a. Change and stability in G O O S E , G O AT and F O O T : Back vowel dy‐ namics in Carlisle English. English Language and Linguistics (First view). References 136 Jansen, Sandra. 2017b. External and internal factors in a levelling process - Prevocalic (r) in Carlisle English. In Beal Joan and Hancil, Sylvie (eds.) Perspectives on northern Englishes. Berlin: Mouton de Gruyter. 111-134. Jansen, Sandra. 2018. Peripheral communities and innovation - Changes in the G O O S E vowel in a West Cumbrian town. In Natalie Braber and Sandra Jansen (eds.) Sociolin‐ guistics in England. Basingstoke: Palgrave Macmillan. 297—332. Jansen, Sandra. submitted. A crossdialectal perspective on a ubiquitous sound change: The case of T-glottaling in two Cumbrian communities. Jansen, Sandra. in preparation. Social changes and levelling processes in a peripheral community: Prevocalic / r/ in Maryport English. In Sandra Jansen and Lucia Siebers (eds.) Processes of change: Studies in Late Modern and Present-Day English. Amsterdam, New York: John Benjamins. Jansen, Sandra and Thorsten Brato. 2008. ‘You used to gerri’ in yer fishnet now you only gerri’ in yer nightdress’: Regional and supraregional accents in English rock songs. Paper presented at Methods in Dialectology XIII. Leeds: University of Leeds. Johnson, Daniel Ezra 2009. Getting off the GoldVarb standard: Introducing Rbrul for mixed-effects variable rule analysis. Language and Linguistics Compass 3: 359-383. Johnstone, Barbara and Scott F. Kiesling. 2008. Indexicality and experience: Variation and identity in Pittsburgh. Journal of Sociolinguistics 12: 5-33. Jones, Daniel. 1932. An outline of English phonetics, 3 rd edition. Leipzig: Teubner. Jones, Lucy. 2012. Dyke/ Girl: Language and identities in a lesbian group. London: Palgrave Macmillan. Jones, Mark J. and Carmen Llamas. 2008. Fricated realisations of / t/ in Dublin and Mid‐ dlesbrough English: An acoustic analysis of plosive frication and surface fricative contrasts. English Language and Linguistics 12: 419-443. Kamata, Miho. 2006. A socio-phonetic study of the D R E S S , T R A P and S T R U T vowels in London English. Leeds Working Papers in Linguistics and Phonetics 11. Kendall, Tyler. 2009. Vowel capture script. Available online: http: / / ncslaap.lib.ncsu.edu/ tools/ Kerswill, Paul. 1996. Children, adolescents, and language change. Language Variation and Change 8: 177-202. Kerswill, Paul. 2003. Dialect levelling and geographical diffusion in British English. In David Britain and Jenny Cheshire (eds.) Social dialectology. Amsterdam: John Benja‐ mins. 223-243. Kerswill, Paul. 2004. Koineization and accommodation. In J. K. Chambers, Peter Trudgill and Natalie Schilling-Estes (eds.) The Handbook of language variation and change. Ox‐ ford: Blackwell. 669-702. References 137 Kerswill, Paul. 2009. Language and social class. In Jonathan Culpeper, Francis Katamba and Paul Kerswill (eds.) English Language: Description, variation and context. Basing‐ stoke: Palgrave Macmillan. 358-372. Kerswill, Paul and Ann Williams. 2002. Salience as an explanatory factor in language change: evidence from dialect levelling in urban England. In Mari C. Jones and Edith Esch (eds.) Language change: The interplay of internal, external, and extra-linguistic factors. Berlin, New York: Mouton de Gruyter. 81-110. Labov, William. 1963. The social motivation of a sound change. Word 19: 273-309. Labov, William. 1972a. Language in the inner city. Philadelphia: University of Pennsyl‐ vania Press. Labov, William. 1972b. Sociolinguistic patterns. Philadelphia: University of Pennsylvania Press. Labov, William. 1984. Field methods of the project on linguistic change and variation. In John Baugh and Joel Sherzer (eds.) Language in use. Englewood Cliffs: Prentice Hall. 28-53. Labov, William. 1990. The intersection of sex and social class in the course of linguistic change. Language Variation and Change 2: 205-254. Labov, William. 1994. Principles of linguistic change: Linguistic factors. Oxford: Blackwell. Labov, William. 2001. Principles of linguistic change: Social factors. Malden, Mass.: Black‐ well. Labov, William. 2003. Pursuing the cascade model. In David Britain and Jenny Cheshire (eds.) Social dialectology. Amsterdam: John Benjamins. 9-22. Labov, William. 2006. The social stratification of English in New York City, 2 nd edition. Cambridge: Cambridge University Press. Leach, Hannah, Kevin Watson and Ksenia Gnevsheva. 2016. Perceptual dialectology in northern England: Accent recognition, geographical proximity and cultural promi‐ nence. Journal of Sociolinguistics 20: 192-211. Lenz, Alexandra. 2010. Zum Salienzbegriff und zum Nachweis salienter Merkmale. In Christina A. Anders, Markus Hundt and Alexander Lasch (eds.) Perceptual dialec‐ tology: Neue Wege der Dialektologie, 2nd edition. Berlin: Walter de Gruyter. 89-110. Levon, Erez and Sue Fox. 2014. Social salience and the sociolinguistic monitor. Journal of English Linguistics 42: 185-217. Llamas, Carmen. 1998. Language variation and innovation in Middlesbrough: a pilot study. Leeds Working Papers in Linguistics and Phonetics 6: 97-114. Llamas, Carmen. 2001a. Language variation and innovation in Teesside English. PhD thesis. Leeds: University of Leeds. Llamas, Carmen. 2001b. The sociolinguistic profiling of (r) in Middlesbrough English. In Hans van de Velde and Roeland van Hout (eds.) ‘r-atics: Sociolinguistics, phonetic and References 138 phonological characteristics of / r/ (Special Issue of Études et Travaux). Brussels: ILVP. 123-140. Llamas, Carmen. 2007. "A place between places": Language and identities in a border town. Language in Society 36: 579-604. Llamas, Carmen, Dominic Watt and Daniel Ezra Johnson. 2009. Linguistic accommoda‐ tion and the salience of national identity markers in a border town. Journal of Lan‐ guage and Social Psychology 28: 381-407. MacRaild, Donald M. 1998. Culture, conflict, and migration: The Irish in Victorian Cum‐ bria. Liverpool: Liverpool University Press. Maddieson, Ian. 2011. Presence of uncommon consonants. In Matthew Dryer and Martin Haspelmath (eds.) The World Atlas of Language Structures Online. Munich: Max Planck Digital Library. Chapter 19. Mallinson, Christine and Robin Dodsworth. 2010. Revisiting the need for new approaches to social class in variationist sociolinguistics. Sociolinguistic Studies 3: 253-278. Marsden, Sharon. 2006. A sociophonetic study of labiodental / r/ in Leeds. Leeds Working Papers in Linguistics and Phonetics 11: 153-172. Mathisen, Anne G. 1999. Sandwell, West Midlands: Ambiguous perspectives on gender patterns and models of change. In Paul Foulkes and Gerard Docherty (eds.) Urban voices: Accent studies in the British Isles. London: Edward Arnold. 107-123. McCarthy, Mike. 1993. Carlisle: History and guide. Stroud: Alan Sutton Publishing. Mees, Inger M. 1987. Glottal stop as a prestigious feature in Cardiff English. English World-Wide 8: 25-39. Mees, Inger M. and Beverly Collins. 1999. Cardiff: a real-time study of glottalization. In Paul Foulkes and Gerard Docherty (eds.) Urban voices: Accent studies in the British Isles. London: Edward Arnold. 185-202. Mesthrie, Rajend. 2010. Socio-phonetics and social change: Deracialisation of the G O O S E vowel in South African English. Journal of Sociolinguistics 14: 3-33. Meyerhoff, Miriam. 2011. Introducing sociolinguistics, 2nd edn. London: Routledge. Milroy, James. 1996. A current change in British English: Variation in (TH) in Derby. Newcastle and Durham Working Papers in Linguistics 4: 213-222. Milroy, James and Lesley Milroy. 1985. Linguistic change, social network and speaker innovation. Journal of Linguistics 21: 339-384. Milroy, James, Lesley Milroy, Sue Hartley and David Walshaw. 1994. Glottal stops and Tyneside glottalization: Competing patterns of variation and change in British Eng‐ lish. Language Variation and Change 6: 327-357. Milroy, Lesley. 1987. Language and social networks. Oxford: Blackwell. Milroy, Lesley and Matthew Gordon. 2008. Sociolinguistics: Method and interpretation, 2 nd edition. Malden: Wiley-Blackwell. References 139 Montgomery, Chris. 2006. Northern English dialects: a perceptual approach. PhD thesis. Sheffield: Sheffield University. Montgomery, Chris. 2012. The effect of proximity in perceptual dialectology. Journal of Sociolinguistics 16: 638-668. Moore, Emma. 2010. The interaction between social category and social practice: Ex‐ plaining was/ were variation. Language Variation and Change 22: 347-371. Mufwene, Salikoko S., John Rickford, John Baugh and Guy Bailey (eds.) (1998). Af‐ rican-American English: Structure, history, and use. London, New York: Routledge. Newbrook, Mark. 1999. West Wirral: norms, self-reports and usage. In Paul Foulkes and Gerard Docherty (eds.) Urban voices: Accent studies in the British Isles. London: Edward Arnold. 90-106. Orton, Harold and Eugen Dieth. 1962. Survey of English dialects. Leeds: Arnold. Piercy, Caroline. 2012. A transatlantic cross-dialectal comparison of non-prevocalic / r/ . In University of Pennsylvania Working Papers in Linguistics 18(2). Article 10. Przedlacka, Joanna. 2001. Estuary English and RP: Some recent findings. Studia Anglica Posnaniensia 36: 35-50. Reany, P. H. 1927. A grammar of the dialect of Penrith (Cumberland). Manchester: Man‐ chester University Press. Reid, Euen. 1978. Social and stylistic variation in the speech of children: Some evidence from Edinburgh. In Peter Trudgill (ed.) Sociolinguistic patterns in British English. London: Arnold. 158-171. Rydland, K. 1978. Vowel systems and lexical-phonemic patterns in south-east Cumbria: A study in structural dialectology. Bergen: University of Bergen. Sangster, Catherine M. 2001. Lenition of alveolar stops in Liverpool English. Journal of Sociolinguistics 5: 401-412. Sankoff, Gillian. 2006. Age: Apparent time and real time. In Keith Brown (ed.) Encyclo‐ pedia of Language & Linguistics. Elsevier. 110-116. Schleef, Erik. 2013. Glottal replacement of / t/ in two British capitals: Effects of word frequency and morphological compositionality. Language Variation and Change 25: 201-223. Schwarz, Christian, Helmut Spiekermann and Tobias Streck. 2011. Primäre und sekun‐ däre Dialektmerkmale. Empirische Befunde aus Dialekten und Standardvarietäten. In Heinz-Dieter Pohl (ed.) Akten der 10. Arbeitstagung für bayerisch-österreichische Dia‐ lektologie. Wien: Praesens. 355-377. Shepherd, Margaret E. 2011. Across the oceans: Emigration from Cumberland and West‐ morland before 1914. Carlisle: Bookcase. Shorrocks, Graham. 1998. A grammar of the dialect of the Bolton area: Volume 1: Intro‐ duction; Phonology. Frankfurt am Main: Lang. References 140 Siebers, Lucia. 2007. Investigating Black South African English: The case of Xhosa English. PhD thesis. Essen: University of Duisburg-Essen. SkyNews. 2010. Prince William And Kate's First Interview Since Getting Engaged. Avail‐ able online: http: / / www.youtube.com/ watch? v=NK3ODM5S0Lg (09 May 2012). Smith, Jennifer and Sophie Holmes-Elliott. 2017. The unstoppable glottal: Tracking rapid change in an iconic British variable. English Language and Linguistics 9: 1-33. Smith, Kenneth. 1984. Carlisle. Clapham: Dalesman Books. Steele, Hazel. 2008. Mechanisms, motivations and outcomes of change in Morley (Leeds) English. PhD thesis. York: University of York. Stoddard, Jana, Clive Upton and J.D.A Widdowson. 1999. Sheffield dialect in the 1990s: revisiting the concept of NORMs. In Paul Foulkes and Gerard Docherty (eds.) Urban voices: Accent studies in the British Isles. London: Edward Arnold. 72-89. Stuart-Smith, Jane. 1999a. Glasgow: Accent and voice quality. In Paul Foulkes and Gerard Docherty (eds.) Urban voices: Accent studies in the British Isles. London: Edward Arnold. 181-204. Stuart-Smith, Jane. 1999b. Glottals past and present: A study of T-glottaling in Glaswe‐ gian. Leeds Studies in English 30: 181-204. Stuart-Smith, Jane. 2008. Scottish English: Phonology. In Bernd Kortmann and Clive Upton (eds.) Handbook of varieties of English: The British Isles. Berlin: Mouton de Gruyter. 48-70. Stuart-Smith, Jane and Claire Timmins. 2010. The role of the individual in language var‐ iation and change. In Carmen Llamas and Dominic Watt (eds.) Language and identi‐ ties. Edinburgh: Edinburgh University Press. 39-54. Stuart-Smith, Jane, Claire Timmins and Fiona Tweedie. 2007. 'Talkin' Jockney'? Variation and change in Glaswegian accent. Journal of Sociolinguistics 11: 221-260. Sullivan, Jeremiah. 1857. Cumberland & Westmorland, ancient and modern; the people, dialect, superstitions and customs. London: Whittaker and co. Tagliamonte, Sali. 2006. Analysing sociolinguistic variation. Cambridge: Cambridge Uni‐ versity Press. Thomas, Erik. 2004. Instrumental phonetics. In J. K. Chambers, Peter Trudgill and Natalie Schilling-Estes (eds.) The Handbook of language variation and change. Oxford: Black‐ well. 168-200. Thomas, Erik. 2007. Sociophonetics. In Robert Bayley and Ceil Lucas (eds.) Sociolinguistic variation: Theories. methods, and applications. Cambridge: Cambridge University Press. 215-233. Thomas, Erik and Tyler Kendall. 2007. NORM: The vowel normalization and plotting suite. Torgersen, Eivind. 1997. Some phonological innovations in south-east British English. MA dissertation. Bergen: University of Bergen. References 141 Torgersen, Eivind and Paul Kerswill. 2004. Internal and external motivation in phonetic change: Dialect levelling outcomes for an English vowel shift. Journal of Sociolinguis‐ tics 8: 23-53. Towill, Sydney. 1991. A history of Carlisle. Chichester: Phillimore. Towill, Sydney. 1996. Georgian and Victorian Carlisle: Life, society and industry. Preston: Carnegie. Trudgill, Peter. 1972. Sex, covert prestige and linguistic change in the urban British Eng‐ lish of Norwich. Language in Society 1: 179-195. Trudgill, Peter. 1974. The social differentiation of English in Norwich. Cambridge: Cam‐ bridge University Press. Trudgill, Peter. 1986. Dialects in contact. New York: Basil Blackwell. Trudgill, Peter. 1988. Norwich revisited: Recent changes in an English urban dialect. English World-Wide 9: 33-49. Trudgill, Peter. 1994. Dialects. London: Routledge. Trudgill, Peter. 1999a. Norwich: endogenous and exogenous linguistic change. In Paul Foulkes and Gerard Docherty (eds.) Urban voices: Accent studies in the British Isles. London: Edward Arnold. 124-140. Trudgill, Peter. 1999b. The dialects of England, 2nd edn. Oxford: Blackwell. Trudgill, Peter. 2002. Sociolinguistic variation and change. Edinburgh: Edinburgh Univer‐ sity Press. Upton, Clive. 2008. Synopsis: phonological variation in the British Isles. In Bernd Kort‐ mann and Clive Upton (eds.) Handbook of varieties of English: The British Isles. Berlin: Mouton de Gruyter. 269-282. Wakelin, Martyn F. 1972. English dialects: An introduction. London: Athlone Press. Wales, Katie. 2008. Northern English: A social and cultural history. Cambridge: Cambridge University Press. Warburton, Jasemine. 2017. Intra-regional variation in north east England. Poster pre‐ sented at UK Language Variation and Change 11. Cardiff: Cardiff University. Watson, Kevin. 2006. Phonological resistance and innovation in the North-West of Eng‐ land. English Today 22: 55-61. Watt, Dominic. 1998. Variation and change in the vowel system of Tyneside English. PhD thesis. Newcastle: Newcastle University. Watt, Dominic. 2000. Phonetic parallels between the close-mid vowels of Tyneside Eng‐ lish: Are they internally or externally motivated? Language Variation and Change 12: 69-101. Watt, Dominic. 2002. ‘I don’t speak with a Geordie accent, I speak, like, the Northern accent’: Contact-induced levelling in the Tyneside vowel system. Journal of Sociolin‐ guistics 6: 44-63. References 142 Watt, Dominic and Anne Fabricius. 2003. Evaluation of a technique for improving the mapping of multiple speakers' vowel spaces in the F1~F2 plane. Leeds Working Papers in Linguistics and Phonetics 9: 159-173. Watt, Dominic, Anne Fabricius and Tyler Kendall. 2010. More on vowels: plotting and normalization. In Marianna Di Paolo and Melcah Yaeger-Dror (eds.) Sociophonetics: A student's guide. London: Routledge. 107-118. Watt, Dominic, Carmen Llamas and Daniel Ezra Johnson. 2010. Levels of linguistic ac‐ commodation across a national border. Journal of English Linguistics 38: 270-289. Watt, Dominic and Lesley Milroy. 1999. Patterns of variation and change in three New‐ castle vowels: is this dialect levelling? In Paul Foulkes and Gerard Docherty (eds.) Urban voices: Accent studies in the British Isles. London: Edward Arnold. 25-46. Watt, Dominic and Jennifer Tillotson. 2001. A spectrographic analysis of vowel fronting in Bradford English. English World-Wide 22: 269-303. Watts, Emma L. 2005. Mobility induced dialect contact: A sociolinguistic investigation of speech variation in Wilmslow, Cheshire. PhD thesis. Colchester: University of Essex. Wells, John C. 1982. Accents of English: The British Isles. Cambridge: Cambridge University Press. West, Helen F. 2009. 'I'm not Geordie! I'm not actually anything! ' Convergent and divergent trends: Dialect levelling and the struggle for identity in a South Durham New-Town. MA dissertation. Edinburgh: University of Edinburgh. West, Helen F. 2013. A town between dialects: Accent levelling, psycho-social orientation and identity in Merseyside, UK. In Peter Auer, Javier Caro Reina and Göz Kaufmann (eds.) Language Variation - European Perspectives IV: Selected papers from the Sixth International Conference on Language Variation in Europe (ICLaVE 6), Freiburg, June 2011. Amsterdam: John Benjamins. 247-266. Williams, Ann and Paul Kerswill. 1999. Dialect levelling: Change and continuity in Milton Keynes, Reading and Hull. In Paul Foulkes and Gerard Docherty (eds.) Urban voices: Accent studies in the British Isles. London: Edward Arnold. 141-162. Wodak, Ruth and Gertraud Benke. 2017. Gender as a sociolinguistic variable: New per‐ spectives on variation studies. In Florian Coulmas (ed.) The Handbook of sociolinguis‐ tics. Malden: Blackwell. 127-150. Wolfram, Walt and Natalie Schilling-Estes. 2006. American English: Dialects and varia‐ tion, 2 nd edition. Malden, MA: Blackwell. Wright, Joseph. 1898-1905. The English dialect dictionary. Wright, Peter. 1977 (1978 edition). How they talk in Carlisle. The Journal of the Lakeland Dialect Society: 7-15. Wright, Peter. 1979. Cumbrian dialect. Clapham: Daleman Books. www.visionofbritain.org.uk. References 143 Acknowledgements PhD theses are never the product of only one person’s work. Many people are involved in the process and I would like to thank them here. First and foremost, I would like to thank my supervisor Raymond Hickey (Chief) for the many discussions about language change in Cumbria and his continuous support. I also fondly remember the various Wednesday afternoon office parties. I would also like to thank Evelyn Ziegler for agreeing to become my second examiner at very short notice. Mercedes Durham read parts of the thesis and provided invaluable feedback. I would also like to thank Rainer Schulze for being so very patient with me when life was too hectic to focus on the reworking of the thesis. A special thanks goes to my family who do not understand what I am doing as a linguist but who supported me in every possible way. Last but not least, I need to thank the people in Carlisle who agreed to talk to me and tell me all about their city. Without you, there would be no thesis. Thank you. List of Figures Figure 2.1: Accommodation as described by Giles and Powesland (1975; cited in Trudgill 1986) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13 Figure 2.2: Accommodation process as described by Hickey (2000) . . . . . . 13 Figure 3.1: The geographical position of Carlisle in Britain (d-maps.com) 24 Figure 5.1.1: Historical relationship between G O O S E and F OOT (taken from Wells (1982: 198)) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44 Figure 5.1.2: Scale for fronting proposed by Mesthrie (2010: 12) for data normalised with mW&F . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49 Figure 5.1.3: Fronting in the G OO S E vowel according to the speakers’ age and sex . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53 Figure 5.1.4: Vowel plot of a 60-year-old MC male speaker (13M60MC) . . 54 Figure 5.1.5: Vowel plot of a 23-year-old MC female speaker (2F23MC) . . 55 Figure 5.1.6: Fronting of the G O O S E vowel according to environment . . . . . 56 Figure 5.2.1: Percentage of [t] and [ʔ] for (T) according to age . . . . . . . . . . 70 Figure 5.2.2: Percentage of variants of (T) according to age . . . . . . . . . . . . 72 Figure 5.2.3: Percentage of variants of (T) according to sex . . . . . . . . . . . . . 73 Figure 5.2.4: Percentage of variants of (T) according to social class . . . . . . 74 Figure 5.2.5: Percentage of [ʔ] in intervocalic position according to age group, sex and social class . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75 Figure 5.2.6: Percentage of variants of (T) according to style . . . . . . . . . . . 77 Figure 5.2.7: Percentage of variants of (T) according to environment . . . . . 78 Figure 5.3.1: Percentage of variants of (R) according to age . . . . . . . . . . . . 91 Figure 5.3.2: Percentage of variants of (R) according to sex . . . . . . . . . . . . . 93 Figure 5.3.3: Percentage of variants of (R) according to age and sex . . . . . 94 Figure 5.3.4: Percentage of variants of (R) according to social class . . . . . . 94 Figure 5.3.5: Percentage of variants of (R) according to age, sex and social class . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95 Figure 5.3.6: Percentage of variants of (R) according to style . . . . . . . . . . . 98 Figure 5.3.7: Percentage of variants of (R) according to environment . . . . . 99 Figure 5.3.8: Preceding environment in #C_V position when [ɾ] is used (N = 119) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101 Figure 5.4.1: Percentage of variants of (TH) according to age . . . . . . . . . . . 114 Figure 5.4.2: Percentage of variants of (TH) according to sex . . . . . . . . . . . 116 Figure 5.4.3: Percentage of variants of (TH) according to social class . . . . . 117 Figure 5.4.4: Use of for (TH) according to individual speaker and environment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 119 List of Figures 146 List of Tables Table 4.1: Education levels . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33 Table 4.2: Occupation types . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34 Table 4.3: SEC groups for this study (adapted from Hall 2008: 69) . . . . . . . 34 Table 4.4: Final sample used in this study . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37 Table 5.1.1: Distribution of lexical subsets of G O O S E in this study . . . . . . . 46 Table 5.1.2: Average F2 values of G O O S E represented by the normalised average mW&F score for each speaker and its approximate position in the vowel chart according to Mesthrie (2010) . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50 Table 5.1.3: Rbrul output results for G O O S E based on the F2 measurements of all G O O S E tokens . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52 Table 5.1.4: Distribution of average F2 values across approximate positions according to Mesthrie (2010) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56 Table 5.1.5: Rbrul output results for G OAT based on the F1 measurements of all G OAT tokens . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57 Table 5.1.6: Rbrul output results for G OAT based on the F2 measurements of all G OAT tokens . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58 Table 5.2.1: Absolute numbers and percentages for each variant of (T) according to social class, sex and age . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69 Table 5.2.2: Rbrul output results for the use of alveolar stops for all speakers 70 Table 5.2.3: Rbrul output for the use of glottal stops for all speakers . . . . . 71 Table 5.2.4: Absolute numbers and percentages for each variant of (T) according to style . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76 Table 5.2.5: Absolute numbers and percentages for each variant of (T) according to environment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77 Table 5.2.6: Frequency order of [ʔ] for sex categories . . . . . . . . . . . . . . . . . 80 Table 5.3.1: Absolute numbers and percentages for each variant of (R) according to social class, sex and age . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86 Table 5.3.2: Rbrul output for use of approximants for all speakers . . . . . . . 87 Table 5.3.3: Rbrul output for use of taps for all speakers . . . . . . . . . . . . . . . 88 Table 5.3.4: Rbrul output for use of labiodental / r/ for all speakers . . . . . . 89 Table 5.3.5: Rbrul output for use of zero for all speakers . . . . . . . . . . . . . . . 89 Table 5.3.6: Absolute numbers and percentages for each variant of (R) according to style . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97 Table 5.3.7: Absolute numbers and percentages for each variant of (R) according to environment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99 Table 5.4.1: Absolute numbers and percentages for each variant of (TH) according to social class, sex and age . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 110 Table 5.4.2: Rbrul output for use of voiceless dental fricatives for all speakers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111 Table 5.4.3: Rbrul output for use of labiodental fricatives for all speakers . 111 Table 5.4.4: Absolute numbers and percentages for each variant of (DH) according to social class, sex and age . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112 Table 5.4.5: Absolute numbers and percentages of [f] for (TH) . . . . . . . . . . 114 Table 5.4.6: Absolute numbers and percentages for each variant of (TH) according to style . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 117 Table 5.4.7: Absolute numbers and percentages for each variant of (TH) according to environment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 118 Table 6.1: Overview of diffusional features found in Carlisle English (check marks in brackets mean that the feature was used twice at the most) . . . . 123 Table 6.2: Groups who are leading a change . . . . . . . . . . . . . . . . . . . . . . . . . 126 List of Tables 148