ArticlePDF Available

‘Don’t even [/f/h]ink aboot it’: An ethnographic investigation of social meaning, social identity and () variation in Glasgow

Authors:

Abstract and Figures

As a relatively new phenomenon in the phonology of Scottish English, TH-fronting has surprised sociolinguists by its rapid spread in the urban heartlands of Scotland. While attempts have been made to understand and model the influence of lexical effects, media effects and frequency effects, far less understood is the role of social identity. Using data collected as part of an ethnographic study of a high school in the south side of Glasgow, Scotland, this article addresses this gap in the literature by considering how TH-fronting is patterned across three all-male, working-class, adolescent Communities of Practice, and how this innovative variant is integrated within a system of the more established variants [] and [h]. Drawing on recent work on linguistic variation and social meaning, the article also explores some of the social meanings of (), particularly those variants which previous research has reported as being associated with ‘toughness’, and suggests how these meanings are utilised in speakers’ construction of social identity.
Content may be subject to copyright.
is is a contribution from English World-Wide 35:1
© 2014. John Benjamins Publishing Company
is electronic le may not be altered in any way.
e author(s) of this article is/are permitted to use this PDF le to generate printed copies to
be used by way of oprints, for their personal use only.
Permission is granted by the publishers to post this le on a closed server which is accessible
to members (students and sta) only of the author’s/s’ institute, it is not permitted to post
this PDF on the open internet.
For any other use of this material prior written permission should be obtained from the
publishers or through the Copyright Clearance Center (for USA: www.copyright.com).
Please contact rights@benjamins.nl or consult our website: www.benjamins.com
Tables of Contents, abstracts and guidelines are available at www.benjamins.com
John Benjamins Publishing Company
English World-Wide : (), –.  ./eww...law
 – / - – © John Benjamins Publishing Company
‘Don’t even [θ/f/h]ink aboot it
An ethnographic investigation of social meaning,
social identity and (θ) variation in Glasgow*
Robert Lawson
Birmingham City University
As a relatively new phenomenon in the phonology of Scottish English, -
fronting has surprised sociolinguists by its rapid spread in the urban heartlands
of Scotland. While attempts have been made to understand and model the inu-
ence of lexical eects, media eects and frequency eects, far less understood is
the role of social identity. Using data collected as part of an ethnographic study
of a high school in the south side of Glasgow, Scotland, this article addresses
this gap in the literature by considering how -fronting is patterned across
three all-male, working-class, adolescent Communities of Practice, and how this
innovative variant is integrated within a system of the more established variants
[θ] and [h]. Drawing on recent work on linguistic variation and social meaning,
the article also explores some of the social meanings of (θ), particularly those
variants which previous research has reported as being associated with ‘tough-
ness’, and suggests how these meanings are utilised in speakers’ construction of
social identity.
Keywords: Glaswegian, sociophonetics, (θ), adolescent male language use,
linguistic variation, Community of Practice, ethnography
* I am grateful to Lynn Clark for her feedback on an earlier dra of this article and to Costas
Gabrielatos, Daniel Johnson, Miriam Meyerho, Jane Stuart-Smith and James Walker for their
help and advice on using Rbrul. Questions and comments from audiences at the University
of Pittsburgh and conversations with Alana DeLoge, Fawn Drucker, Andy Jeske, Barbara
Johnstone, Scott Kiesling, Claude Mauk and Abdesalam Soudi helped me clarify some of the
thoughts presented here. Lastly, particular thanks go to Edgar W. Schneider for his detailed
editorial guidance and to the two anonymous reviewers for their suggestions and comments.
© 2014. John Benjamins Publishing Company
All rights reserved
(θ) variation in Glasgow 
. Introduction
e increase in -fronting1 in Scotland (where /θ/ is realised as [f] in words like
think and through, Wells 1982: 328), has been one of the major surprises in recent
sociolinguistic research on Scottish speech communities for a number of reasons.
First, -fronting is not a traditional phonetic feature in Scottish English, espe-
cially in word-initial position, where speakers have the choice of the standard vari-
ant [θ] or, in certain classes of words, the non-standard variant [h]. Second, the
leaders of -fronting appear to be non-mobile working-class adolescents rather
than middle-class, loosely-tied, mobile speakers (cf. Milroy 1987). ird, although
-fronting is a typical feature of London English (Wells 1982; for a general dis-
cussion of -fronting in the UK see Foulkes and Docherty 1999: 17; Kerswill
2003), the widespread adoption of the non-standard variant [f] by working-class
speakers in Scotland is a relatively recent trend, contravening well-established
models about how linguistic change is propagated (cf. the geographical diusion
model, Kerswill 2003: 239–40 and the dialect contact model, Robinson 2005: 189;
Stuart-Smith and Timmins 2006: 171).
e increase of -fronting in Scotland has consequently raised a number
of questions about how and why it is spreading, and in particular, why a feature
strongly associated with London varieties of English is being used by speakers
who have little to no contact with this variety. Stuart-Smith and Timmins (2010)
argue that the inuence of the media, including television programmes such as
EastEnders (a London based soap drama), is one possible reason for the increase of
-fronting in Scotland (and Glasgow specically), while other work on -front-
ing in Scotland has focused on understanding the inuence of lexical category
(Stuart-Smith and Timmins 2006), lexical frequency (Clark and Trousdale 2009),
dialect contact (Brato 2012) and phonological structure (Schleef and Ramsammy
2013).
But while much has been written on media eects, lexical eects and fre-
quency eects, very little has been said about the inuence of social identity on
a speaker’s pattern of variation. is is surprising given the robust research lit-
erature which shows that a speaker’s social identity is oen one of the key pre-
dictors of linguistic variation (Eckert 2000; Moore 2003; Lawson 2011, although
. Although Wells (1982: 328) uses the term ‘-fronting’ to refer to “the replacement of the
dental fricatives [θ, ð] with the labiodentals [f] and [v] respectively” (Wells 1982: 328), I follow
Stuart-Smith and Timmins (2006) who use the term to refer only to the replacement of the
voiceless dental fricative [θ] with the voiceless labio-dental fricative [f]. e term ‘-fronting’
is reserved for the replacement of [v] for [ð] and is not discussed here.
© 2014. John Benjamins Publishing Company
All rights reserved
 Robert Lawson
see Trudgill 2001, 2008 for a critique of the centrality of identity in treatments of
sociolinguistic variation).
Answering the call made by Stuart-Smith and Timmins (2010: 53) for a more
ethnographic approach in order to understand the intersection of identity catego-
ries and -fronting in Glasgow, this article addresses this gap by oering an eth-
nographically-informed account of linguistic variation among three adolescent,
all-male, working-class Communities of Practice in the south side of Glasgow.
Using data collected over the course of a long-term study in a high school named
Banister Academy,2 I outline how the current picture of (θ) in Glasgow can be
augmented by ne-grained ethnographic description which uncovers locally-
grounded social distinctions. As such, this article contributes to our understand-
ing of how variants of (θ) are distributed across an apparently homogenous group
of speakers, furthering our knowledge of the social meaning of -fronting in
Glasgow.
In the next section of the article, I discuss the background of (θ) and -
fronting in Scotland before moving on in Section 3 to consider the relationship
between linguistic variation and social meaning. In Section 4, I discuss the rel-
evance of the Community of Practice model within ethnographic sociolinguistics,
followed by an outline of the Communities of Practice encountered in Banister
Academy in Section 5. In Section 6, I discuss the methodology used in the study,
while Section 7 presents the results of the quantitative analysis of (θ). Section 8
brings together the qualitative and quantitative material and discusses how vari-
ants of (θ) are used in the construction of social identity, and particularly the re-
lationship between non-standard variants and specic conceptualisations of mas-
culinity.
. (θ) and -fronting in Scotland
Traditionally, Glaswegian (and other varieties of Scottish English more generally)
has two main variants of the variable (θ):3 the standard dental fricative variant [θ]
and the local, non-standard glottal fricative variant [h]. In terms of linguistic con-
straints, [h] is principally found in word-initial position (although it can also be
used in word-medial position, Clark and Trousdale 2009: 41; Lawson 2009: 102),
and is restricted to a particular subset of high-frequency lexemes, namely think
. All names given in the eldwork (school name, participants, etc.) are pseudonyms.
. In some parts of Scotland, such as Fife in the east, an additional variant [∫] exists, usually
before /r/ (e.g. three). is variant, however, is rarely attested in Glasgow (Clark and Trousdale
2009: 39).
© 2014. John Benjamins Publishing Company
All rights reserved
(θ) variation in Glasgow 
and thing and their associated lexical derivatives (such as thinking, thinks, things
etc., Clark and Trousdale 2009: 39). [θ], on the other hand, has no such lexical
restrictions. ese variants also have a clear class distribution such that [θ] is pri-
marily used by middle-class speakers (Stuart-Smith and Timmins 2006) while [h]
is primarily used by working-class speakers.
Although [θ] and [h] are well-established in Glasgow, however, a new variant
[f] appears to have gained considerable ground over the last 20 years. Indeed, the
rapid spread of [f] has, as mentioned above, been a central focus in recent phono-
logical and sociolinguistic research in Scotland.
e rst formal mention of -fronting in Glasgow can be found in Macafee
(1983: 34, Fn. 26), who notes that even though the variant is not evidenced in
her dataset, there is some anecdotal evidence that young speakers occasionally
use the ‘Cockney form’ [f]. Stuart-Smith, Timmins and Tweedie (2007: 254) sug-
gest that the variant can be traced back even earlier, to at least the mid 1950s,
and there are some examples of [f] reported in folk poetry and songs from the
1960s onwards (one notable example includes the line “they’re aw Caicks up
oor close”).4
To date, the most systematic investigation on -fronting in the city has been
the work of Jane Stuart-Smith and colleagues, using two corpora of conversational
data (one collected in 1997 and the other collected in 2003). eir analysis shows
that since 1997, [f] has steadily increased in use, particularly among working-class
adolescents (Stuart-Smith, Timmins and Tweedie 2007: 236). One reason for this
is that unlike [h], [f ] is more productive, occurring in word-initial position (e.g.
three, [fri]), intervocalic position (e.g. gothic, [gɔfɪk]), and word-nal position (e.g.
both, [bof]). Consequently, the variant has a far greater potential for uptake across
the lexicon. [h], on the other hand, is primarily restricted to a small subset of
high-frequency lexemes, namely the think/thing set and their associated lexical
derivatives (Stuart-Smith and Timmins 2006).
It appears to be the case, however, that these high-frequency lexemes ‘slow
downthe spread of -fronting. In their analysis of frequency eects in data col-
lected as part of a linguistic ethnography of a pipe band in Fife (named the West of
Fife Highland Pipe Band, or WFHPB), Clark and Trousdale (2009: 49) comment
that although speakers are more likely to adopt the innovative form [f] in higher
frequency words than lower frequency words, the high-frequency words of the
/ lexical set actually resist innovation due to their entrenched status
within the lexicon.
. ‘ey’re all Catholics in our alleyway’. I am grateful to one of the anonymous reviewers for
this example.
© 2014. John Benjamins Publishing Company
All rights reserved
 Robert Lawson
Beyond lexical and frequency eects, recent work has also touched on how
speakers use -fronting as a resource in the sociolinguistic construction of iden-
tity. For example, Clark and Trousdale (2009: 51) highlight how social identity is a
signicant predictor of the use of [f] in their data, pointing out that;
[e] factor group ‘community of practice/friendship group membership’ sub-
stantially outranks all other constraints on the variation. In other words, there is a
very strong correlation between the use of the labiodental fricative and member-
ship in a particular social group in this community.
Similarly, the analysis by Stuart-Smith, Timmins and Tweedie (2007) intimates
that social identity is a signicant factor in understanding how -fronting is pat-
terned across urban adolescent speaker communities, although their discussion
lacks the social detail aorded by ethnographic eldwork.
Understanding how -fronting is implicated in the construction of social
identity is intimately tied to understanding what social meanings surround -
fronting in Scotland more generally. As Clark (2009: 154) highlights, despite the fact
that -fronting is strongly associated with varieties of Southern British English
(particularly London varieties), adolescent Scottish speakers do not appear to be
adopting [f] as a way of marking alignment with English nationality. Stuart-Smith,
Timmins and Tweedie (2007) also oer limited evidence that Scottish adolescents
orientate positively towards England, arguing instead that adolescent speakers in
Glasgow “seem to be using all possible linguistic resources to construct identities
which are as anti-middle-class, and anti-establishment, as possible” (Stuart-Smith,
Timmins and Tweedie 2007: 251). Relatedly, Clark (2009: 155) suggests that be-
cause -fronting is more associated with working-class speakers, one of the so-
cial meanings of [f] is ‘rough, at least within WFHPB.
As Trudgill (1972) and Kiesling (1998) point out, non-standard variants are
typically associated with physicality and male power, so it is perhaps no surprise
that as a non-standard variant, [f] can index ‘roughness. Taking these discussions
as a point of departure, I consider in this article why [f] might mean ‘rough, how
far this meaning of ‘roughness’ is enacted across other adolescent communities,
and how -fronted variants interact with other variants of (θ) in Glasgow. Before
turning to these points, however, I briey consider in the next section the extant
literature on variation, social meaning and social identity.
© 2014. John Benjamins Publishing Company
All rights reserved
(θ) variation in Glasgow 
. You are what you speak?: Variation, social meaning and social identity
In the early years of sociolinguistic research, variation was viewed as a simple re-
ection of social dierences. For example, Trudgill’s analysis of (ing) in Norwich
(1972) showed that male speakers used higher rates of the alveolar nasal variant
[n] while female speakers used higher rates of the velar nasal variant [ŋ], irre-
spective of social class. Adopting a ‘rst waveperspective (Eckert 2005), Trudgill
theorised that speakers used specic patterns of variation because they were male
or female (see also Levon 2010: 11). While rst wave studies oered robust correla-
tions between linguistic variation and social groupings (e.g. Labov 1972; Macaulay
1977), they were challenged on the basis of their binary treatment of gender, the
limited attention on the intersection of class and gender, their focus on more mac-
ro-level patterns of variation, and their view of style as primarily controlled by a
speaker’s attention to their speech.
In ‘second wave’ studies, sociolinguists began to focus on more local speech
communities (e.g. Milroy 1987), but it was only in ‘third wave’ studies where so-
ciolinguists started treating variation as a constituent part of social identity in-
stead of seeing linguistic variation simply as a reection of social dierences. As
Eckert (2005: 30) points out, the change from rst wave to third wave approach-
es “[moved] the study of variation o in a new direction. Rather than dening
variation in terms of the speakers who use variables, [third wave studies seek]
the meanings that motivate particular variable performances”. More specically,
Eckert’s quote highlights the need for sociolinguists to move beyond a straightfor-
ward description of who uses which variants and instead to see style as ‘persona
constructionrather than ‘attention to speech’ and thus tackle how linguistic vari-
ables acquire ‘social meaning’.
In this vein of work, one group instrumental to developing a theoretical frame-
work at the intersection of linguistic variation, social meaning and social identity
was the California Style Collective (1993). Drawing on anthropological work on
subcultures (Hebdige 1979), one of their key arguments was that ‘meaning mak-
ing’ typically occurs through a process of ‘bricolage, where “a speaker chooses bits
of linguistic practice from various sources and recombines them in order to create
a ‘style’ or identity” (Kiesling 1998: 71). e process of bricolage relies, in part, on
the notion of indexicality, that is “the creation of semiotic links between linguistic
forms and social meanings” (Bucholtz and Hall 2005: 594).
Of course, the social meaning of a linguistic variable is also dependent on the
kinds of speakers who use it and the context in which it is used (Eckert 2002: 5),
but since such insights are dicult to obtain without some way of understand-
ing the kinds of social positionings and stances adopted by participants, sociolin-
guists working within the third wave paradigm have relied on the deployment of
© 2014. John Benjamins Publishing Company
All rights reserved
 Robert Lawson
ethnographic methodology to uncover the social meaning of a variable, moving
beyond the global and situating linguistic practice in the local sphere. Importantly,
ethnography privileges a ‘bottom-up’ approach, rather than the ‘top-down’ ap-
proach characteristic of many early quantitative sociolinguistic studies. is ap-
proach also necessitated the introduction of new framework, the ‘Community of
Practice’.
. Communities of Practice and ‘ird Wave’ sociolinguistics
Although sociolinguists are primarily concerned with the processes of linguistic
variation and change, Mendoza-Denton (2008: 3) argues that “[linguists] must
look at language by looking beyond language, we must look holistically at the life-
world of the people with whom we work and investigate the richness of practices
that are inextricably tied to language. Since this perspective is not characteristic
of rst wave studies, third wave studies required new ways of investigating so-
cial practices and showing how these practices were linked to the language use.
Consequently, one of the key developments of the third wave model was the in-
troduction of the Community of Practice (CofP hereaer), incorporated into
sociolinguistics in an inuential article on language and gender by Eckert and
McConnell-Ginet (1992: 464) in which they dened a CofP as
an aggregate of people who come together around mutual engagement in an en-
deavor. Ways of doing things, ways of talking, beliefs, values, power relations — in
short, practices — emerge in the course of this mutual endeavor.
Extending these ideas, Eckert and McConnell-Ginet (1999) and Eckert (2005: 17)
argue that engagement with, and production of, practices is how speakers con-
struct their social identities, and linguistic variation is viewed as a key social prac-
tice through which speakers show their allegiance and membership of particular
social groups while simultaneously constructing their identities as members of
that group.
Linguistic variation is, however, only part of the story of understanding
how speakers construct their social identities, and Mendoza-Denton’s comment
highlights that it is not enough for sociolinguists to focus wholly on variation in
tackling this endeavour. Indeed, work which focuses on social practices beyond
language has shown how central non-linguistic practices are in mapping out the
complex processes of identity construction, and in a traditional quantitative study,
there is a danger of overlooking social distinctions which do not emerge at the
level of language and assuming that speakers in a particular social context belong
to the same speech community. By adopting a CofP approach, we can go beyond
© 2014. John Benjamins Publishing Company
All rights reserved
(θ) variation in Glasgow 
a straightforward quantitative description of linguistic variation and integrate a
range of social information in our pursuit of better understanding the sociolin-
guistic construction of identity. is perspective has, however, been thus far rela-
tively neglected in discussions of linguistic variation in Glasgow.
. Communities of Practice in Banister Academy
e data considered in this article were collected during a three-year longitudi-
nal ethnography of a high school in the south side of Glasgow named Banister
Academy between 2005 and 2007, and in the analysis below, I focus on data col-
lected from three CofPs which I named the “Alternative, “Sports”, and “Ned”
CofPs. e speakers in each of the CofPs were between 14 and 15 years old at the
beginning of the eldwork, and each CofP consisted of three speakers.5
Membership of a particular CofP was decided through a combination of
self-identication, other-identication and ethnographic observation, along with
identifying patterns of shared social practice. Since CofP membership is uid and
changeable, triangulation is a useful way of capturing a specic constellation of
CofP membership at a particular time. Due to the dynamic nature of CofP mem-
bership and the vagaries of ethnographic eldwork, however, not all speakers are
represented in each year (see Lawson 2011: 233 for a breakdown of CofP mem-
bership by year). Some problems arose with the lack of opportunities to gain ac-
cess to particular CofPs and speakers, while other speakers were unavailable for
recording for a number of reasons (e.g. not attending school). e practical eects
of these diculties meant the speakers in each year of data sometimes vary. For
example, membership of the Alternative CofP changed the most radically, with
some members from Year 1 leaving the CofP entirely, while the Sports CofP was
relatively stable throughout the eldwork. Rather than conate speakers across
each year of data, however, and consequently run the risk of misrepresenting the
dynamic membership of each CofP, the decision was taken to keep each year of
analysis separate.
e members of the Alternative CofP were some of the rst pupils I met at
Banister Academy and it was immediately obvious that a number of their social
practices fell outside what would be considered ‘mainstream. For example, they
rejected almost all mainstream sports, such as football and rugby, and preferred a
range of ‘extreme sports, including BMX riding and American wrestling. Similarly,
their musical tastes were more in-line with rock and metal bands such as Nirvana,
. Although this is a relatively small number of speakers per CofP, this is not uncommon in
ethnographically-informed sociolinguistic studies (e.g. Moore 2003; Mendoza-Denton 2008).
© 2014. John Benjamins Publishing Company
All rights reserved
 Robert Lawson
Cradle of Filth, Iron Maiden, Metallica, and Slipknot, and knowledge of this genre
of music was a particularly valuable social commodity. eir clothing choices re-
ected these social practices, with many members opting to wear t-shirts branded
with their favourite band or wrestling icon, or black leather coats and biker boots
instead of a regular sports jacket and trainers preferred by other CofPs.
In contrast, the Sports CofP were marked out by their engagement with a
range of mainstream sports, primarily football6 and rugby. Football, and to a lesser
extent rugby, were the sports around which many of the Sports CofP members
structured their days, both inside and outside Banister Academy. Football in par-
ticular was a key activity pursued every break time and lunchtime, with certain
venues in the school playground carefully guarded (or contested) for use by the
group. As a result of their daily sporting activities, the Sports CofP members opted
to wear trainers, school jumper with a t-shirt, or a white shirt, either with or with-
out the school tie. None of the members of the Sports CofP wore the school blazer,
instead wearing a regular jacket over their uniform. e Sports CofP also invested
time watching mainstream television programmes such as EastEnders which was
a regular topic of conversation.
Before discussing the last CofP, it is worthwhile outlining what the term ‘ned’
means within Glaswegian society more generally, particularly for readers unfa-
miliar with social labels in Glasgow. e term typically refers to a working-class
adolescent male who wears a tracksuit, Burberry branded clothing (recognisable
by its plaid design), a baseball cap, white sports trainers, gold sovereign rings,
and a Berghaus ‘Mera Peak’ jacket (an expensive hiking jacket) who is assumed
to be involved in a range of anti-social, criminal and/or age-restricted behaviours,
including vandalism, assault, the, drinking, smoking and drug-taking. As such,
the stereotypical picture of a Glaswegian ned is almost wholly negative, exploited
in comedic media representations (e.g. the BBC comedy show Chewin’ the Fat
contains a segment entitled ‘e Two Neds’ in which two urban adolescent males
engage in a variety of anti-social behaviour against two middle-class older males).
My main aim in using the term, however, is not to further the discourse of
social marginalisation and stigmatisation which typically surrounds this group,
but rather to highlight the fact that the group was marked out as dierent by their
peers. is distinction was partially based on the assumption that they were in-
volved in a range of anti-social and criminal behaviours, but there were also other,
more innocuous, reasons, including the belief that such individuals misbehaved
more in class. Indeed, the Ned CofP were typically involved in the local subcul-
ture, including skipping school, participating in age-restricted activities such as
. My rst ever interaction with the members of the Sports CofP was through being invited to
play a game of football (soccer) one lunchtime.
© 2014. John Benjamins Publishing Company
All rights reserved
(θ) variation in Glasgow 
smoking and drinking, and low-level crime such as petty the and minor vandal-
ism, but there was no evidence of them being involved in more serious criminal
activity such as assaults or muggings (a few were, however, involved in isolated
instances of gang-related violence). In terms of dress, the members of the Ned
CofP only adopted a few of the stereotypical practices. For example, one member
wore a Mera Peak jacket and several of them wore gold sovereign rings, yet the de-
ployment of these specic practices was not uniform across the entire CofP, thus,
it is dicult to talk of a ‘ned’ identity as a homogenous entity, in contrast to the
dominant picture held by many Glaswegians.
. Data and analysis
In an auditory analysis of approximately 15 hours of conversational data, 550
tokens of word initial (θ) were identied and marked up using (version
5.1.09). Given the lexical constraints outlined above, word initial (θ) was separated
out into two main patterns in the descriptive analysis (Sec. 7). Pattern I covers
those words which can take either [θ], [f], and [h] (e.g. think, thing and associated
lexical derivatives), and Pattern II covers those words which can take only [θ] or
[f] (e.g. through, throw). Clark and Trousdale (2009) take Pattern I words to also
cover (θ) in word-medial position (e.g. everything), but in this article I focus only
on (θ) in word-initial position.
e data were subject to a binomial mixed-model logistic regression analy-
sis using the add-on Rbrul for the programming package R (Johnson 2009),
an approach which has grown considerably in recent sociolinguistic research
(Drummond 2011; Schleef, Meyerho and Clark 2011; Schleef and Ramsammy
2013). Unlike traditional ‘xed-eect’ models used in sociolinguistics (such as
Varbrul and GoldVarb), Rbrul accounts for both ‘xed’ eects (such as male / fe-
male, young / old, CofP membership) and ‘randomeects (that is, factors which
are not normally replicable, such as the individual speakers recruited for a study).
As Drummond (2011: 294) points out, “by including individual speaker as a ran-
dom eect, the model is able to account for the fact that some speakers may favour
a particular variant to a greater or lesser degree than their relevant xed factors
would predict” (see also Johnson 2009: 365).
An additional benet of Rbrul is the way in which the program deals with the
independence of observations in a dataset, an approach which is very dierent to
that utilised in older statistical soware (e.g. GoldVarb). In GoldVarb, each token
is treated as though it is a single observation, unrelated to previous or future ob-
servations. As Johnson (2009: 363) notes, however, in most quantitative linguistic
analysis, tokens are not independent since they are grouped according to which
© 2014. John Benjamins Publishing Company
All rights reserved
 Robert Lawson
speaker produced them. In programs such as GoldVarb, there are two ways of
dealing with this issue. e rst is to leave out the factor group of ‘speaker’. In this
approach, however, the inuence of external variables, like gender or age, will be
dramatically overestimated. e second option is to include ‘speaker’ as a factor
group, yet this can lead to a model in which external eects are underestimated
and those factors which might be the best t are excluded. Johnson (2009: 364)
outlines the diculties in adding ‘speaker’ as a xed eect in a statistical model,
but because Rbrul can include ‘speaker’ as a random eect, it is better equipped
to deal with the issue of determining which eects are more inuential, external
eects or inter-speaker variation.
Johnson (2009) also highlights the fact that most sociolinguistic datasets are
unbalanced and tokens are not evenly distributed across speakers. Since this can
impact on the ‘t’ of a statistical model, it is necessary to use a statistical method
which can ‘balance out’ the unevenness inherent in most sociolinguistic data-
sets. With regard to the data presented in this article, it should be noted that
there was indeed an unequal distribution of tokens across speakers. Since the
data were collected from spontaneous conversation, there was no way to ensure
that every speaker would produce an equal (or near equal) number of (θ) tokens.
Consequently, some speakers used a high number of tokens with (θ) while oth-
ers used a relatively low number. Nevertheless, Rbrul is equipped to deal with
such imbalances. Additionally, Rbrul deals better with Type I errors, where a
chance eect is identied as a real eect (in contrast to a Type II error which
is when a real eect goes unidentied). Consequently, the program is able to
capture external eects, but only when they are strong enough to rise above
the inter-speaker variation” (Johnson 2009: 365). Lastly, Rbrul is generally more
suited to large-scale sociolinguistic studies and thus larger ‘pots’ of tokens, and
while it would have been preferable to have analysed Pattern I and Pattern II
words together, the fact that [f] is an innovative non-standard variant while [h]
is a traditional non-standard variant means that it is dicult to conate them in
one statistical model.
Consequently, models were run on Pattern I and Pattern II datasets separately
following the breakdown outlined below:
1. Pattern I (A) — Standard variant vs. non-standard variants (i.e. [θ] vs. [f, h])
2. Pattern I (B) — Non-standard variant I vs. non-standard variant II (i.e. [f] vs.
[h])
3. Pattern II — Standard variant vs. non-standard variant (i.e. [θ] vs. [f])
e data were also coded for the linguistic and social constraints outlined in
Table 1.
© 2014. John Benjamins Publishing Company
All rights reserved
(θ) variation in Glasgow 
Preceding and following phonological category were coded in this way in
order to discover whether variants of (θ) were inuenced by the manner of ar-
ticulation of the preceding and/or following segment, and which variant would
be more likely before a vowel or a pause. Preceding and following vowel was not
coded for front/back since previous research had shown this not to be a signi-
cant predictor (Clark and Trousdale 2009: 43). Because the nature of the dataset
was restricted from the outset, a number of factors which have been included in
other statistical treatments of (θ) could not be included in the Banister Academy
data. For example, since all the speakers are male, it was not possible to code for
‘male’ and ‘female’, as we would normally expect in a traditional sociolinguistic
analysis. Similarly, class was also controlled across the sample, as were region and
age. Moreover, by focusing only on word initial (θ), all tokens have /θ/ in the onset
position, so there is no need to code for ‘onset’ and ‘codaposition (e.g. think vs.
both), nor can we code for the position of (θ) in the word (initial, medial or nal).
So although the number of predictive factors here seem rather small (compared to,
for example, Clark and Trousdale [2009: 43–4] or Schleef and Ramsammy 2013),
this is primarily to do with the structure of the dataset than any underlying disad-
vantage of the coding process.
. Results of analysis of (θ)
In the following section, I present the results of the analysis of (θ) across all three
CofPs. I rst outline the general patterns of variation of (θ) in Pattern I and Pattern
II in each year, followed by the results of the regression analysis for each pattern. In
Section 8, I consider the results in light of the ethnographic eldwork conducted
in Banister Academy to develop a better picture of the social meaning of variants
of (θ) and how these variants are implicated in the construction of social identity
in Glasgow.
Table 1. Independent social and linguistic variables used in statistical model
Categorical variables Factor levels
CofP membership Sports ~ Alternative ~ Ned
Preceding class Vowel ~ Pause ~ Obstruent ~ Sonorant
Following class Vowel ~ Pause ~ Obstruent ~ Sonorant
# of Syllable Monosyllabic ~ Multisyllabic
Topic Television ~ Area ~ Age-restricted ~ event Sports ~ Friends ~ Dating
~ School ~ Fighting ~ Filler ~ ‘Neds’ ~ Arguing ~ Undetermined
Random eects Speaker and Word
© 2014. John Benjamins Publishing Company
All rights reserved
 Robert Lawson
. Pattern I results: Year 1
For Pattern I (θ) in Year 1, there is a higher use of the standard variant [θ] by the
Alternative CofP speakers. e Alternative CofP speakers also tend to use a lower
rate of [h] than the Sports CofP speakers, although Jack is something of an outlier
here and is categorical in his use of [h], outstripping all other speakers in the Year
1 sample,7 while Andrew is the only Alternative CofP speaker to use [f].
In the Sports CofP, there appears to be a stronger tendency for [h], although
Phil uses a slightly lower rate of this variant, patterning more with Andrew
(Alternative CofP) than with the two other members of the Sports CofP. Phil and
Nathan also use several tokens of [f], while Mark is the only speaker to use [θ].
0%
20%
40%
60%
80%
100%
Andrew
Jack
Neil
Mark
Nathan
Phil
Percentage
Speakers
n = 127
Sports
Alternative
(θ)
[f]
[h]
Figure 1. Word Initial (θ): Pattern I among all speakers in Year 1
Two main regression analyses were run; the rst with ‘standard’ versus ‘non-stan-
dard’ as the dependent variable (so [θ] ~ [f, h]) with ‘non-standard’ as the appli-
cation value, and the second with [f ] versus [h] with [h] as the application value.
Four independent variables were entered into the model (‘Community of Practice
membership, ‘preceding phonological class, ‘topic’ and ‘number of syllables’),
and both ‘word’ and ‘speaker’ were entered as random eects. Following phono-
logical class was not entered since in Pattern I words the following phonological
. Interestingly, Lawson (2011) showed that Jack’s  vowel was also distinctly dierent from
the other Alternative CofP speakers, suggesting a low level of involvement within this CofP.
Indeed, in Year 2 and Year 3 of the eldwork, Jack gave up membership of the Alternative CofP
and began socialising with another group in Banister Academy, suggesting that his peripheral
status within the Alternative CofP precluded acquisition of their patterns of variation.
© 2014. John Benjamins Publishing Company
All rights reserved
(θ) variation in Glasgow 
class is always a vowel. In both statistical models, however, none of the predictors
emerged as signicant. is is not surprising given the relatively disjointed picture
of variation presented here.
. Patterns II results: Year 1
For Pattern II (θ) in Year 1, there is a clear division according to CofP membership,
with the Sports CofP categorical in their use of [f]. e results for the Alternative
CofP show that they use both the standard [θ] and non-standard [f], with Neil the
leader for [θ]. ese results, in combination with the results for Pattern I (θ) sug-
gests that, overall, the Alternative CofP speakers are more standard. It is important
to note, however, that even though the Sports CofP speakers are categorical in us-
ing [f] in the Pattern II context, they only produced 23 tokens in total. Andrew, on
the other hand, produced 45 tokens in total, with a far wider lexical range than the
Sports CofP speakers. Despite this, however, for those words which are common
across both the Alternative and Sports CofPs (namely three and through), we have
evidence that the Sports CofP speakers never use the standard variant, unlike the
Alternative CofP speakers.
0%
20%
40%
60%
80%
100%
Percentage
n = 97
Andrew
Jack
Neil
Mark
Nathan
Phil
Speakers
Sports
Alternative
(θ)
[f]
Figure 2. Word Initial (θ): Pattern II among all speakers in Year 1
Only one regression analysis was run with [θ] versus [f] (since [h] is not per-
missible in Pattern II words) and [f ] as the application value. Five independent
variables were entered into the model (‘Community of Practice membership, ‘pre-
ceding phonological class’, ‘following phonological class’, ‘topic’ and ‘number of
syllables’), and both ‘word’ and ‘speaker’ were entered as random eects. For this
model, CofP emerged as the only signicant factor (Table 2).
© 2014. John Benjamins Publishing Company
All rights reserved
 Robert Lawson
From the model above, the log odds and factor weights show that speakers
from the Sports CofP prefer the variant [f]. at CofP is signicant in this model
aligns with the results outlined in the descriptive analysis above, where [f] was
categorical among Sports CofP speakers.
. Pattern I results: Year 2
Moving on to consider word initial (θ) in Year 2, we can see that the results alluded
to in Year 1 become more developed. For example, in the Alternative CofP we see
that Kevin and Mathew have a similar rate of the standard variant [θ], and all three
speakers have similar rates for [f]. For [h], however, Peter uses a relatively high
rate, while Kevin and Mathew use a far lower rate. As was found in Year 1, it ap-
pears that the Alternative CofP generally use a higher rate of the standard variant
and relatively lower rates of the non-standard variants.
Continuing the pattern of variation established in Year 1, the main variant in
Pattern I words in the Sports CofP is [h], with all three speakers using a very high
rate of this variant, and it is categorical for Nathan. Compared to the Alternative
CofP, the Sports CofP have low rates of [f] and [θ], with Phil the only speaker to
use [f], and both Mark and Phil using low rates of [θ].
While the low number of tokens produced by the Ned CofP means that we
must be cautious about interpreting these results, we can nevertheless oer some
tentative conclusions, with the main nding being that all three speakers are cat-
egorical users of [h].
As with the Pattern I data in Year 1, two regression analyses were conducted.
Unlike the results for Pattern I data in Year 1, however, both regression analyses
returned CofP as a signicant factor in the model (Table 3 and Table 4).
As the log odds and factor weights in Table 3 show, speakers from the Ned
CofP prefer non-standard variants, while the Alternative CofP speakers are most
likely to use the standard variant [θ]. More specically, as Table 4 shows, the
Ned CofP speakers strongly prefer [h] rather than [f] (also demonstrated in the
Table 2. Statistical model for word initial (θ) Year 1: Pattern II
Application value:
‘non-standard’
Factor Tokens Log-odds Factor weight
CofP membership
p < 0.01
‘Neds’ 27 11.133 > 0.999
Sports 66 −3.794 0.022
Alternative 62 −7.339 0.001
Not signicant Topic, preceding phonological class, number of syllables
Model Deviance = 100.675 df = 9 intercept = 7.503 mean = 0.761
© 2014. John Benjamins Publishing Company
All rights reserved
(θ) variation in Glasgow 
descriptive analysis above), while the Sports CofP speakers are more likely to use
[h] rather than [f] (albeit at a lower rate), and the Alternative CofP speakers more
likely to use [f] rather than [h].
(θ)
[f]
[h]
0%
20%
40%
60%
80%
100%
Kevin
Mathew
Peter
Mark
Nathan
Phil
Danny
Max
Noah
Percentage
Speakers
n = 156
Sports Alt Ned
Figure 3. Word Initial (θ): Pattern I among all speakers in Year 2
Table 3. Statistical model for word initial (θ) Year 2: Pattern I (A)
Application value: [h] Factor Tokens Log-odds Factor weight
CofP membership
p < 0.01
‘Neds’ 27 12.312 > 0.999
Sports 64 −4.817 0.008
Alternative 28 −7.495 0.001
Not signicant Topic, preceding phonological class, number of syllables
Model Deviance = 77.111 df = 9 intercept = 7.166 mean = 0.840
Table 4. Statistical model for word initial (θ) Year 2: Pattern I (B)
Application value: [f] Factor Tokens Log-odds Factor weight
CofP membership
p < 0.05
Sports 23 8.546 > 0.999
Alternative 75 −8.546 < 0.001
Not signicant Topic, preceding phonological class, following phonological class,
number of syllables
Model Deviance = 47.826 df = 9 intercept = 10.581 mean = 0.908
© 2014. John Benjamins Publishing Company
All rights reserved
 Robert Lawson
. Pattern II results: Year 2
For Pattern II words in Year 2, all three speakers in the Alternative CofP again
use both the standard and non-standard variants. More specically, Peter uses the
highest rate of [f] and the lowest rate of [θ], while Mathew and Kevin use relatively
similar rates of both variants (Figure 4). e tendency towards [f] continues in the
Sports CofP, with all three speakers using high rates of the non-standard variant.
Mark is the only speaker who displays a marked use of [θ] which accounts for over
35% of his overall variation, while [f] is categorical for Phil and almost categorical
for Nathan, who uses only one token of [θ] in the word three. e results for Year 2
are slightly dierent from what we might expect given the results in Year 1 where
all three speakers were categorical in their use of [f].
For the Ned CofP, the low token count means that again, the results should be
taken cautiously. Nevertheless, the main tendency in Pattern II is for the Ned CofP
speakers to use [f]. Indeed, both Danny and Noah have categorical [f], but the
results for Max show that he uses both [θ] and [f]. Max, however, only produces
ve tokens of (θ) across the entire dataset, meaning that even small variations in
his variant choice will have dramatic eects on the nal results (and on the results
of the statistical analysis). Indeed, over the course of his interview, Max produced
only two tokens of [θ] which accounted for 40% of his variation (for the word
three), although he used both [f] and [θ] in this lexical item.
e regression analyses did not return CofP (or any other predictor entered
into the model) as signicant, a result which can be attributed to Max’s pattern of
Kevin
Mathew
Peter
Mark
Nathan
Phil
Danny
Max
Noah
Speakers
Sports Alt Ned
0%
20%
40%
60%
80%
100%
Percentage
n = 170
(θ)
[f]
Figure 4. Word Initial (θ): Pattern II among all speakers in Year 2
© 2014. John Benjamins Publishing Company
All rights reserved
(θ) variation in Glasgow 
variation. Indeed, when Max was removed from the model, CofP approached sig-
nicance (p = 0.0596), while the omission of the Ned CofP from the model showed
that CofP also came close to signicance (p = 0.0617). Both regressions showed
that with the omission of Max, CofP trended towards being a signicant predictor
in the model, although it just failed to reach signicance.
. Discussion
Since previous research on (θ) in Glasgow conates Pattern I and Pattern II words
(e.g. Stuart-Smith and Timmins 2006; Stuart-Smith, Timmins and Tweedie 2007),
it is dicult to directly compare previous research with the Banister Academy
data. Nevertheless, we can conrm existing trends on the progress of -fronting
in Glasgow on a more general level (Table 5).
Table 5. Comparison of main phonetic variants for (θ) in word initial position in the
1997 Corpus, 2003 Corpus, and Banister Academy data
Variants 1997 2003 2009 (Banister Academy)
% n % n % n
[θ] 28.3 88 15.5 194 16.55 91
[h] 43.4 151 49.1 710 46.18 254
[f] 26.2 102 34.5 409 37.27 205
Total 100 341 100 1313 100 550
A comparison of the Banister Academy results with those reported by the Glasgow
Speech Project shows that, generally, the use of [f] is increasing at the expense of
[θ], while [h] remains relatively stable. is, however, is only part of the story,
since when we look at the local patterns of variation in Banister Academy, we see
that even within an apparently homogenous group of working-class adolescent
male speakers, there are ne-grained patterns of (θ) variation.
e results show that the Alternative CofP is generally the most standard, the
Ned CofP speakers are the most non-standard, and the Sports CofP falls some-
where in-between. Following the third-wave approach, however, it is not enough
to simply describe the variable patterning across these three CofPs, and in order
to add an extra layer of analytical description, it is important to account for why
dierent CofPs use variants of (θ) in dierent ways and the social meaning of
variants of (θ) within Banister Academy (and by extension, Glasgow more gener-
ally). Taking as a point of departure Clark’s claim that [f] indexes ‘rough’ and the
claim by Stuart-Smith, Timmins and Tweedie (2007: 251–3) that [f] is involved in
a complicated process of locally based language ideologies in which -fronted
© 2014. John Benjamins Publishing Company
All rights reserved
 Robert Lawson
variants are part of an ‘anti-establishment’ stance, we can begin to piece together
how far these meanings are developed across the CofPs in Banister Academy, the
extent to which the potential social meanings of [f] interact with the other vari-
ants of (θ) in the city, and how these variants might be exploited by speakers in the
construction of specic social identities.
. ‘Hink yer hard aye?’ Roughness and toughness in Glasgow
Glasgow is a post-industrial city with a high concentration of working-class com-
munities, and as such, it is reasonable to assume that the views of the young men
in Banister Academy about what constitutes ‘real’ masculinity would be inuenced
in part by the historical and socioeconomic reality of the city in which they live
(Johnson and McIvor 2004, 2007). Indeed, one specic conceptualisation of mas-
culinity involves the evocation of a range of behavioural traits which would be
considered as hegemonically masculine, including physical toughness, bravery,
and a willingness to ght, crystallised most obviously in the concept of the ‘hard
man, a working class male identity structured around the actual or potential de-
ployment of such practices (see also Lawson 2013 for a more detailed discussion
of this idea).
In Glasgow (and Scotland more generally), the non-standard variant [h] is
strongly identied as a social class marker, and specically as a marker of working-
class speech. As noted above, variants usually acquire meaning by virtue of the
people who use them, and since the concept of the ‘hard man’ (and by extension,
notions of ‘roughnessand ‘toughness’) is most strongly associated with working-
class communities (Trudgill 1972), an indirect indexical chain emerges whereby
non-standard variants come to be associated with such meanings. For example,
the following tweet (collected in May 2012), draws attention to the association of
Glaswegian with violence and aggression, an association which is regularly ex-
ploited in media representations of Glasgow (cf. e Big Man’ from the BBC com-
edy show Chewin’ the Fat).
@Joey7Barton hink yer hard aye? Mon doon the pitches in glasgow sunday nights
n ill break both yer legs then yer jaw ya scouse scumbag8
In the tweet, the dense cluster of Glaswegian features (such as use of the 
vowel in doon, the lexical item aye, the contracted form mon for come on, and the
use of [h] in think), in conjunction with imagery of ghting, aggression, and foot-
ball violence, all work together to create a picture of ‘tough’ masculinity.
. ‘ink you’re hard, aye? Come on down to the football (soccer) pitches in Glasgow Sunday
nights and I’ll break both your legs then your jaw, you Scouse scumbag’.
© 2014. John Benjamins Publishing Company
All rights reserved
(θ) variation in Glasgow 
To relate these issues back to the Banister Academy data, given that the use of
[h] is most pronounced within the Ned CofP, it is reasonable to ask how far the
use of categorical [h] within this CofP is related to the construction of a particular
masculine ideal of the ‘hard man.
As I noted above in Section 5, the Ned CofP members were engaged in a
number of age-restricted and anti-social practices, including minor gang vio-
lence, alcohol consumption and smoking. While these were social practices other
CofPs almost completely avoided, they were taken to be part and parcel of being
a member of the Ned CofP and part of being a ‘real man’. Indexing this form of
‘tough’ masculinity is not restricted only to visible social practices like ghting and
drinking, but also extends to particular forms of speech (Parker 1992: 146). More
important, however, is the fact that deploying particular linguistic features which
are indexical of ‘toughness’ means that speakers do not have to rely on other, more
socially-risky, practices like ghting to perform ‘tough’ masculinity, particularly in
situations where the outcome of a ght is not guaranteed. Indeed, social practices
like ghting and other violent physical actions are oen viewed as a ‘last resort’ for
the maintenance of ‘tough’ masculinity due to the inherent risks associated with
using such practices (Anderson 1997). us, if a speaker can convince a second
party that they are capable of violent physical action without actually doing so,
the potential social cost is much lower.
Moreover, speakers engaged in the performance of communally-understood
and socially-embedded identities (such as the ‘hard man’) are typically required
to use ‘valid’ linguistic resources in order to avoid social censure, and the use of
non-standard variants is oen a key component of establishing oneself as ‘tough.9
Importantly, the proposed identity construct must match with particular patterns
of linguistic variation, otherwise the performance is viewed as invalid and illegiti-
mate (drawing on the idea of ‘variation rights’, Mendoza-Denton 2008: 252). ese
‘valid’ linguistic resources in the case of the Ned CofP would be those non-stan-
dard variants which avoid any indexical association with the ‘establishment, and
since the local subculture (the culture towards which the Ned CofP orientates) is
in opposition to the ‘establishment culture’ represented by, for example, the school,
it would be unusual for the Ned CofP speakers to use the standard variant [θ]. is
was conrmed to me during an o-tape conversation with a female member of the
. As a piece of ‘anecdata, I encountered this during a shopping trip in Birmingham city centre
where I witnessed a the at one of the markets. During my attempts to intervene and alert the
stall owner, I was intercepted by one of the thief’s associates who enquired where I was going.
Despite my default variety now being Scottish Standard English, I attempted to diuse the situ-
ation by style-switching into a strongly Scots guise, drawing on (I supposed) well-known as-
sociations of Scots as a ‘friendly’ accent. I was promptly informed that I should “drop the hard
man act”.
© 2014. John Benjamins Publishing Company
All rights reserved
 Robert Lawson
Ned CofP who commented that “only posh people say think” (with [θ] rather than
[f] or [h]). For the speaker, [θ] indexes ‘posh, and I would suggest that such an
ideology is common among other members of the Ned CofP. is also aligns with
the comment by Stuart-Smith, Timmins and Tweedie (2007: 253) that standard
variants such as [x] in loch index ‘middle-class-ness’ in Glasgow, so it is no surprise
to see the same ideology extended to other standard variants such as [θ].
[h] also has strong ‘placeconnotations, and given Glasgow’s reputation as a
city of crime and violence (Davies 2007; Braber and Butternt 2008: 24–5), I sug-
gest that [h] is an important way of indexing ‘Glaswegian-ness’, with all the ideo-
logical associations this entails. For the Ned CofP in particular, their association
with Glasgow is an important part of their identities (Lawson 2009: 365–7), and I
argue that their use of [h] is one way of reifying not only their belonging to the city,
but also of exploiting the more indirect ideological associations being Glaswegian
involves (see also Braber and Butternt 2008 for a discussion of local identity and
sound change in Glasgow).
Within those CofPs which do not draw on the idea of the ‘hard man’ to the
same extent, the idea of ‘tough’ masculinity is nevertheless still important, includ-
ing a desire to be viewed as a competent ghter, a tough individual, and physi-
cally strong. If [h] does mean ‘tough’, then the relatively high rate of [h] within
the Sports CofP is understandable. Indeed, additional evidence of this indexical
meaning of [h] can also be found in the fact that not only is Nathan one of the
categorical [h] users in Year 2, but he is also someone who was strongly invested
in the idea of the ‘hard man’ persona and who attempted to present himself to his
peers as someone ‘not to be messed with. His high use of [h] is, I argue, one of the
ways in which his construction of a ‘tough’ masculine identity is enacted.
Lastly, speakers in the Alternative CofP almost completely reject [h], a nding
which ts in with the notion that the Alternative CofP are more closely aligned
with the ‘establishment’ and orientate more positively towards the school than the
Ned CofPs members. For example, many of the Alternative CofP members com-
mented that they enjoyed coming to school and recognised the importance of edu-
cation in securing well-paid employment. It is reasonable to suggest that the use of
the standard variant [θ] is part of this identity construction.
. Finking about [f] and competing social meanings
e picture is somewhat complicated by the appearance of [f], since [f] is neither
‘local’ nor ‘traditional’, yet it remains a robust and productive variant across all
three CofPs, particularly in the Ned CofP. As was noted above, previous research
has suggested that one of the core meanings of [f] is ‘tough’, and in light of the lead-
ers of [f] in Banister Academy (the Ned CofP), this interpretation makes sense.
© 2014. John Benjamins Publishing Company
All rights reserved
(θ) variation in Glasgow 
However, we also have to nd an explanation for the relatively advanced use
of [f] within the Alternative and Sports CofP. It could be that these two groups of
speakers are demonstrating some sort of ‘anti-establishment’ stance and the use
of [f ] is an important way of doing so, particularly since [f] is the only possible
non-standard variant in Pattern II. But the Alternative CofP (and to a lesser extent,
the Sports CofP) did not construct their identities as ‘anti-establishment’ in the
same way as the Ned CofP members did. For example, while the Alternative CofP
speakers tended to reject most mainstream social practices, at the same time, they
were invested in the idea of schooling and had eyes towards skilled careers once
they le school. Similarly, even though the Sports CofP members were more ori-
entated towards the local subculture than members of the Alternative CofP, they
were still invested in education and were never explicitly ‘anti-school’. I argue that
the simultaneous use of [θ] and [f] among the Sports and the Alternative CofPs
is a way for these speakers to straddle the divide between the ‘anti-establishment
and ‘establishment’ cultures; as ways for these speakers to meet a range of conict-
ing social demands. As such, determining issues of social identity has less to do
with focusing on the meaning of one variant in isolation, but rather understanding
how variants pattern with one another across discourse.
We are still, however, le with the problem of competing indexical meanings
of [f] set out in previous research. In order to reconcile the idea that [f] indexes
‘rough’ (Clark 2008: 155), as well as ‘anti-establishment’ and ‘anti-middle class’
(Stuart-Smith, Timmins and Tweedie 2007: 251), I suggest that there is an ongoing
cyclical relationship at work, whereby enacting an ‘anti-establishment’ culture in-
volves a positive orientation towards the idea of ‘tough’ masculinity and that estab-
lishing a sense of ‘toughmasculinity involves developing an ‘anti-establishment’
stance. is is not to say that [f] does not have an indexical eld (Eckert 2008, see
Figure 5), but that the two meanings of ‘toughness’ and ‘anti-establishment’ are
those ones which are the most salient and productive in Banister Academy, to the
extent that they are dialogically bound together and contingent upon one another,
rather than isolated and insulated.
Figure 5. Possible indexical eld for [f] in Banister Academy
© 2014. John Benjamins Publishing Company
All rights reserved
 Robert Lawson
. Conclusions
is article contributes to ongoing discussions regarding the social productivity
of -fronting in Scotland and complements those results found by Stuart-Smith,
Timmins and Tweedie (2007). In particular, by adopting an ethnographic per-
spective on phonetic variation among young men in Glasgow, we have seen how
the patterning of (θ) across the sample is closely related to CofP membership and
how ethnographic methodology can help us illuminate patterns of linguistic dif-
ferentiation within an apparently homogenous group of speakers. Not only does
this research show how a group of speakers who may be nominally dened as
‘working-class’ have subtly dierent, but socially productive, patterns of variation,
the ndings also lend further support to the well-established idea that speakers
closely attend to language variation in the construction of locally-embedded social
identities. Indeed, by considering the question of why working-class male speak-
ers use non-standard variants, we have developed a more nuanced account of the
social meaning of a relatively new variant in Glaswegian, and importantly, how
it interacts with more established variants of (θ). More specically, I have argued
that the use of non-standard variants can be exploited as a low-risk way for speak-
ers to construct an identity of ‘tough’ masculinity, while the use of both standard
and non-standard variants allow speakers to negotiate the ‘grey area’ between
establishmentand ‘subcultural’ contexts. As such, this article augments Stuart-
Smith and Timmins’ point (2007: 255) that “the amplication or exploitation of
[innovative variants by working-class speakers] … is at least partly to be explained
in terms of the construction of specic, locally situated identities which simulta-
neously signal their own identity and dierentiate them from ‘posh’ people. Since
ethnographic eldwork oers a better insight into how these identities actually
emerge in locally-embedded contexts, we can not only add an extra layer of lo-
cal description, but can also move towards a fuller account of the role played by
both ‘old’ and ‘new’ variants of (θ) in the sociolinguistic construction of identity
in Glasgow.
References
Anderson, Elijah. 1997. “Violence and the inner-city street code. In Joan McCord, ed. Violence
and Childhood in the Inner-City. Cambridge: Cambridge University Press, 1–30. DOI:
10.1017/CBO9780511571015.002
Braber, Natalie and Zoe Butternt. 2008. “Local identity and sound change in Glasgow: a pilot
study”. Leeds Working Articles in Linguistics 13: 22–43.
Brato, orsten. 2012. “A sociophonetic study of Aberdeen English: innovation and conserva-
tism”. Ph.D. thesis, University of Gießen.
© 2014. John Benjamins Publishing Company
All rights reserved
(θ) variation in Glasgow 
Bucholtz, Mary and Kira Hall. 2005. “Identity and interaction: a sociocultural linguistic ap-
proach. Discourse Studies 7: 585–614. DOI: 10.1177/1461445605054407
California Style Collective. 1993. “Variation and personal / group style. Paper presented at New
Ways of Analyzing Variation 22, Ottawa.
Clark, Lynn. 2009. “Variation, change and the usage-based approach. Unpublished Ph.D. thesis,
University of Edinburgh.
—— and Graham Trousdale. 2009. “Exploring the role of token frequency in phonologi-
cal change: evidence from -Fronting in east-central Scotland”. English Language and
Linguistics 13: 33–55. DOI: 10.1017/S1360674308002852
Davies, Andrew. 2007. “Glasgow’s ‘Reign of Terror’: street gangs, racketeering and in-
timidation in the 1920s and 1930s”. Contemporary British History 21: 405–27. DOI:
10.1080/13619460601060413
Drummond, Rob. 2011. “Glottal variation in /t/ in non-native English speech: patterns of acqui-
sition. English World-Wide 32: 280–308. DOI: 10.1075/eww.32.3.02dru
Eckert, Penelope. 2000. Linguistic Variation as Social Practice: e Linguistic Construction of
Identity in Belten High. Oxford: Blackwell.
——. 2002. “Constructing meaning in sociolinguistic variation. Paper presented at the Annual
Meeting of the American Anthropological Association, New Orleans, Louisiana.
——. 2005. “Variation, convention, and social meaning”. Paper presented at the Annual
Meeting of the Linguistic Society of America, Oakland, California.
——. 2008. “Variation and the indexical eld”. Journal of Sociolinguistics 12: 453–76. DOI:
10.1111/j.1467-9841.2008.00374.x
—— and Sally McConnell-Ginet. 1992. “ink practically and look locally: language and
gender as community-based practice”. Annual Review of Anthropology 21: 461–90. DOI:
10.1146/annurev.an.21.100192.002333
—— and ———. 1999. “New generalizations and explanations in language and gender re-
search. Language in Society 28:185–201.
Foulkes, Paul and Gerry Docherty. 1999. “Urban voices: an overview”. In Foulkes and Docherty,
eds. Urban Voices: Accent Studies in the British Isles. London: Hodder Arnold, 1–24.
Hebdige, Dick. 1979. Subcultures. London: Methuen. DOI: 10.4324/9780203139943
Johnson, Daniel Ezra. 2009. “Getting o the GoldVarb standard: introducing Rbrul for
mixed-eects variable rule analysis. Language and Linguistics Compass 3: 359–83. DOI:
10.1111/j.1749-818X.2008.00108.x
Johnson, Ronnie and Arthur McIvor. 2004. “Oral history subjectivity and environmental real-
ity: occupational health histories in twentieth-century Scotland”. Osiris 19: 234–49. DOI:
10.1086/649404
—— and ——. 2007. “Narratives from the urban workplace: oral testomonies and the recon-
struction of men’s work in the heavy industries in Glasgow”. In Richard Rodger and Joanna
Herbert, eds. Testimonies of the City: Identity, Community and Change in a Contemporary
Urban World. Aldershot: Ashgate, 23–44.
Kerswill, Paul. 2003. “Dialect levelling and geographical diusion in British English. In
David Britain and Jenny Cheshire, eds. Social Dialectology: In Honour of Peter Trudgill.
Amsterdam, Philadelphia: Benjamins, 223–43.
Kiesling, Scott. 1998. “Men’s identities and sociolinguistic variation: the case of fraternity men.
Journal of Sociolinguistics 2: 69–100. DOI: 10.1111/1467-9481.00031
Labov, William. 1972. Sociolinguistic Patterns. Philadelphia: University of Pennsylvania Press.
© 2014. John Benjamins Publishing Company
All rights reserved
 Robert Lawson
Lawson, Robert. 2009. “Constructions of social identity among adolescent males in Glasgow.
Unpublished Ph.D. thesis, University of Glasgow.
——. 2011. “Patterns of linguistic variation among Glaswegian adolescent males”. Journal of
Sociolinguistics 15: 226–55. DOI: 10.1111/j.1467-9841.2011.00477.x
——. 2013. “e construction of ‘tough’ masculinity: negotiation, alignment and rejection.
Gender and Language 7: 369–395.
Levon, Erez. 2010. Language and the Politics of Sexuality: Lesbians and Gays in Israel. Basingstoke:
Palgrave Macmillan. DOI: 10.1057/9780230281318
Macafee, Caroline. 1983. Glasgow. Amsterdam, Philadelphia: Benjamins.
Macaulay, Ronald. 1977. Language, Social Class, and Education: A Glasgow Study. Edinburgh:
Edinburgh University Press.
Mendoza-Denton, Norma. 2008. Homegirls: Language and Cultural Practices among Latina
Youth Gangs. London: Wiley-Blackwell.
Milroy, Lesley. 1987. Language and Social Networks. 2nd ed. Oxford: Blackwell.
Moore, Emma. 2003. “Learning style and identity: a sociolinguistic analysis of a Bolton high
school”. Unpublished Ph.D. thesis, University of Manchester.
Parker, Howard. 1992. A View from the Boys: A Sociology of Down-Town Adolescents. Aldershot:
Gregg Revivals.
Robinson, Christine. 2005. “Changes in the dialect of Livingston”. Language and Literature 14:
181–93. DOI: 10.1177/0963947005051289
Schleef, Erik, Miriam Meyerho and Lynn Clark. 2011. “Teenagers acquisition of variation: a
comparison of locally-born and migrant teens’ realisation of English (ing) in Edinburgh
and London. English World-Wide 32: 206–36. DOI: 10.1075/eww.32.2.04sch
—— and Michael Ramsammy. 2013. “Labio-dental fronting of /θ/ in London and Edinburgh:
a cross-dialectal study”. English Language and Linguistics 17: 25–54. DOI: 10.1017/
S1360674312000317
Stuart-Smith, Jane and Claire Timmins. 2006. “‘Tell her to shut her moof’: the role of the lexicon
in -fronting in Glaswegian. In Christian Kay, Graham Caie, Carole Hough and Irene
Wotherspoon, eds. e Power of Words: Essays in Lexicography, Lexicology and Semantics.
Amsterdam: Rodopi, 171–183.
—— and ———. 2010. “e role of the individual in language variation and change. In
Carmen Llamas and Dominic Watt, eds. Language and Identities. Edinburgh: Edinburgh
University Press, 39–54.
——, —— and Fiona Tweedie. 2007. “Talkin’ Jockney? Variation and change in Glaswegian
accent”. Journal of Sociolinguistics 11: 221–60. DOI: 10.1111/j.1467-9841.2007.00319.x
Trudgill, Peter. 1972. “Sex, covert prestige and linguistic change in the urban British English of
Norwich. Language and Society 1: 179–95. DOI: 10.1017/S0047404500000488
——. 2001. “On the irrelevance of prestige, stigma and identity in the development of New
Zealand English”. New Zealand English Journal 15: 42–6.
——. 2008. “Colonial dialect contact in the history of European languages: on the irrelevance
of identity to new-dialect formation. Language in Society 37: 241–54.
Wells, John. 1982. Accents of English 1: An Introduction. Cambridge: Cambridge University
Press. DOI: 10.1017/CBO9780511611759
© 2014. John Benjamins Publishing Company
All rights reserved
(θ) variation in Glasgow 
Author’s address
School of English
6th Floor Baker Building
City North Campus
Birmingham City University
Birmingham, B42 2SU
robert.lawson@bcu.ac.uk
© 2014. John Benjamins Publishing Company
All rights reserved
... Third-wave sociolinguistic research has increasingly sought to describe the social meaning of variation, as opposed to identifying broad correlations between variants and macro-level socio-demographic categories (see inter alia Moore 2003;Campbell-Kibler 2010;Eckert 2012;Lawson 2014;Levon 2016;Drummond 2018). Within this line of inquiry, scholars have documented the ways in which features are deployed to convey specific interactional stances and social personae. ...
... Kiesling links these stances to the performance of hegemonic masculinity, arguing that the use of [n] is part of the display of physical power, characterised by dominance and solidarity. Similar themes are also examined in Lawson's (2014) ethnography of male peer groups where he contends that the use of TH-fronting by speakers of the 'Ned' CofP contributes to the development of an 'anti-establishment' stance and their performance of 'tough' masculinity. Here, I suggest that ey is being used in discursively similar ways. ...
Article
Recent accounts of discourse-pragmatic (DP) variation have demonstrated that these features can acquire social indexical meaning. However, in comparison to other linguistic variables, DP features remain underexplored and third-wave perspectives on the topic are limited. In this article, I analyse the distribution, function and social meaning of the ‘attention signals’ – those features which fulfil the explicit function of eliciting the attention of an individual – in just over 35 hours of self-recordings of 25 adolescents collected during a year-long sociolinguistic ethnography of an East London youth group. This leads me to identify an innovative attention signal – ey . Distributional analyses of this feature show that ey is associated with a particular Community of Practice, the self-defined and exclusively male ‘gully’. By examining the discourse junctures at which ey occurs, I argue that this attention signal is most frequently used by speakers to deploy a ‘dominant’ stance. For gully members, this feature is particularly useful as an interpersonal device, where it is used to manage ingroup/outgroup boundaries. Concluding, I link the use of ey and the gully identity to language, ethnicity and masculinity in East London.
... Peters connects his findings in Galway also with more general sociolinguistic patterns, for example the usage of stigmatised local pronunciation features by young male speakers (cf. also Lawson 2014). ...
Chapter
Full-text available
Although a well-established methodological framework in anthropology, criminology, and sociology (Atkinson and Hammersley 2007: 1–10), it has only been over the past 10–15 years that ethnographic methods have seen increased use in (quantitative) sociolinguistics in the UK (see Rampton 2007 for a discussion of linguistics and ethnography in the UK). Scotland in particular has been a key site for research which integrates ethnographic approaches with sociolinguistic investigations of language use and linguistic variation and change in a variety of contexts. This has included research on community organisations (Clark 2009; Clark and Trousdale 2009), Gaelic communities (McEwan-Fujita 2010; Smith-Christmas 2012), national parliaments (Shaw 2009–2011), rural communities (Thomson 2012), schools (Alam 2007; Lambert, Alam and Stuart-Smith 2007; Lawson 2009; Nance 2013), sports clubs (Wilson 2007), and the workplace (Eustace 2012).
Chapter
This chapter investigates the phenomenon of TH-fronting, a change in progress which has rapidly spread across some of the major towns and cities of Britain in the last few decades. In Scotland, TH-fronting most commonly refers to the replacement of the voiceless dental fricative [θ] with the voiceless labiodental fricative [f] (see, for example, Stuart-Smith and Timmins 2006; Clark and Trousdale 2009). Several studies have investigated the social motivations for this phonological change in progress in Scotland (e.g. Robinson 2005; Clark 2009; Lawson, forthcoming), but much less consideration has been given to potential structural motivations. This chapter asks whether structural repetition, or priming, can help to explain variation in TH-fronting in a corpus of vernacular Scots speech.
Article
Full-text available
Previous research has indicated the importance of embodiment in West African emotion lexica. The current study aims to explore the pervasiveness of this cultural script through the analysis of the emotional lexica of two West African languages (Ga and Ewe) from Southern Ghana that have been featured minimally in previous emotion research. The analysis indicated that embodiment was an important cultural script in both affective lexica. However, interpersonal representations of emotions were also present. Further, emotion words in the two languages differed in the more specific loci of emotions.
Article
Full-text available
Drawing on narrative data collected during a three-year ethnography of a Scottish high school, this article examines the construction of working-class adolescent masculinities. More specifically, the analysis focuses on how adolescent male speakers negotiate, reject and align themselves with the hegemonically dominant ideology of 'tough' masculinity, the role socially low-risk discourses of 'tough' masculinity play in interaction, and how speakers integrate a range of discursive strategies which help maintain homosociality when 'tough' masculinity is at stake. I argue that discourses which appear to be about 'being tough' do a great deal more social work than might be expected.
Book
Examining how lesbian and gay Israelis negotiate the linguistic performance of their sexualities and the constraints of Israeli national ideologies, this book broadens current understandings of the uses and effects of variation in language and details the interconnections between language use and sexual, national and political identities.
Chapter
reference = Stuart-Smith, J. and Timmins, C. (2010) The role of the individual in language variation and change. In: Llamas, C. and Watt, D. (eds.) Language and Identities. Edinburgh University Press: Edinburgh, UK, pp. 39-54. ISBN 9780748635764