The authors have declared that no competing interests exist.
Conceived and designed the experiments: AS. Performed the experiments: AS JF RR. Analyzed the data: RS AS. Contributed reagents/materials/analysis tools: RS AS JF RR KC. Wrote the paper: RS AS JF RR KC BW.
This paper presents results from a corpus-based study investigating lexical variation in BSL. An earlier study investigating variation in BSL numeral signs found that younger signers were using a decreasing variety of regionally distinct variants, suggesting that levelling may be taking place. Here, we report findings from a larger investigation looking at regional lexical variants for colours, countries, numbers and UK placenames elicited as part of the BSL Corpus Project. Age, school location and language background were significant predictors of lexical variation, with younger signers using a more levelled variety. This change appears to be happening faster in particular sub-groups of the deaf community (e.g., signers from hearing families). Also, we find that for the names of some UK cities, signers from outside the region use a different sign than those who live in the region.
Variation is an intrinsic part of all languages whether spoken or signed. It is apparent at all levels of language organisation: for example, there are several lexical variants in British Sign Language (BSL) which all mean ‘America’ (see
In an earlier study, sociolinguistic variation and change in BSL numeral signs was investigated. Stamp, Schembri, Fenlon and Rentelis
This paper is organised into five parts: first, we discuss the history of BSL regional variation. Next, we present a brief review of the literature on spoken and signed language lexical variation. We then explain the methodology for the BSL Corpus Project and present the results of this study. Finally, we discuss the findings in relation to other relevant sign language studies and consider their implications for an understanding of variation in BSL.
The 2011 Census for England and Wales reports that between 15,000–20,000 people in the UK use BSL as their main language
Typically language is transmitted from caregiver to child. However, the vast majority of deaf children are born to hearing parents (Mitchell & Karchmer
In recent years, anecdotal claims within the British deaf community suggest that traditional regional variation is in decline, and there is some evidence for these claims, e.g., in numeral signs
In addition, increased mobility and transnational contact have exposed deaf and hearing British people alike to a multitude of languages, dialects and social practices. Research on British English accents has shown how increased mobility within the UK has resulted in dialect levelling
Changes in BSL resulting from the emergence of TV programmes for the deaf community and sign language interpreting on television may have led to an increased preoccupation with political correctness since the 1990s. Signs for foreign countries which portray physical features have sometimes been perceived as ‘racist’ by the hearing non-signing community
Sociolinguistic research has identified the following social factors as providing important insights into the nature of language variation and change in spoken and signed languages: age
Although sociolinguistic investigations have tended to concentrate on variation and change in the phonology and grammar, lexical variables are also an important point of sociolinguistic investigation
In contrast to the focus in contemporary studies in the sociolinguistics of spoken languages on sociophonetic variation, the obvious presence of considerable variation in sign language lexicons has meant that studies of variation in sign language have emphasised lexical variation and change. In contrast to spoken languages, regional and social sign language ‘accents’ have not been described, although some subtle variation in the application of phonological processes to specific sublexical elements does appear to be correlated with region
Theories of regional dialects often implicitly presuppose that there was once a single, uniform language, which diverged until identifiable regional varieties arose, either through spontaneous evolution or language mixing, or both processes
Woll et al.
In contrast to the evidence of considerable traditional lexical variation in BSL, it has been claimed that ASL may have a relatively more standardised lexicon than other documented sign languages
The lexical data in this paper was elicited as part of the BSL Corpus Project. Here we briefly introduce the BSL Corpus Project by outlining the sites of collection, participant characteristics and the methods used in data collection, coding and analysis (for more detail, see
Participants in this research were all deaf with British Sign Language as their main/preferred language. Participants were all aged 16 or over. The University College London (UCL) Research Ethics Committee guidelines state that “young people aged 16–18 with sufficient understanding are able to give their full consent to participate in research independently of their parents and guardians” so no additional consent was obtained from those under 18. It cannot be assumed that members of this language community (deaf BSL users) have fluent or full comprehension of written English. Therefore the comprehensive information statement and consent form in written English which are required by the UCL Research Ethics Committee were translated into British Sign Language by local deaf fieldworkers and deaf researchers working on the project. Questions were clarified in person in BSL and consent obtained in writing. This project including the consent procedure was approved by the UCL Research Ethics Committee (project ID 0864/001). All individuals pictured in this manuscript have given written informed consent to publish their images.
The BSL Corpus Project, which began in 2008, was the first large-scale corpus project to be undertaken for BSL. The aim of the project was to create a corpus of elicited and spontaneous BSL digital video data from deaf native, near-native and early learners of BSL. The project has established an online, open-access video dataset available for researchers and the sign language community
In order to obtain samples of regional variation, data were collected from eight sites across the UK: Belfast, Birmingham, Bristol, Cardiff, Glasgow, London, Manchester and Newcastle. These sites were selected because they are, or were previously, locations of a centralised school for deaf children, and because, as relatively large urban centres, it was assumed that they would provide a sufficiently large deaf community from which to recruit.
Thirty participants were filmed at most sites, although slightly larger samples were collected in Bristol and London, with 32 and 37 participants respectively. In total, 249 deaf individuals were filmed.
Sites | Total | Age | Gender | Ethnicity | Language background | Social class | School location (*some metadata missing) | |||||||
Younger 16–39 | Middle 40–59 | Older 60+ | Female | Male | White | Other | Deaf | Hearing | Working class | Middle class | Local | Non-local | ||
Belfast | 30 | 10 | 12 | 8 | 17 | 13 | 30 | 0 | 7 | 23 | 26 | 4 | 21 | 9 |
Birmingham | 30 | 12 | 9 | 9 | 13 | 17 | 27 | 3 | 12 | 18 | 16 | 14 | 21 | 9 |
Bristol | 32 | 9 | 14 | 9 | 17 | 15 | 30 | 2 | 17 | 15 | 16 | 16 | 16 | 16 |
Cardiff | 30 | 10 | 12 | 8 | 17 | 13 | 28 | 2 | 6 | 24 | 22 | 8 | 10* | 19* |
Glasgow | 30 | 10 | 13 | 7 | 15 | 15 | 27 | 3 | 6 | 24 | 17 | 13 | 20 | 10 |
London | 37 | 8 | 19 | 10 | 17 | 20 | 31 | 6 | 13 | 24 | 15 | 22 | 20* | 16* |
Manchester | 30 | 12 | 6 | 12 | 16 | 14 | 27 | 3 | 8 | 22 | 23 | 7 | 13* | 16* |
Newcastle | 30 | 6 | 11 | 13 | 17 | 13 | 29 | 1 | 7 | 23 | 19 | 11 | 18* | 11* |
TOTAL | 249 | 77 | 96 | 76 | 129 | 120 | 229 | 20 | 76 | 173 | 154 | 95 | 139* | 106* |
The methodology for the BSL Corpus Project was based on two similar large-scale investigations of ASL
In the lexical elicitation task, fieldworkers showed participants PowerPoint slides or flashcards for 102 concepts. Each slide displayed an image of the referent or something associated with it, and the equivalent English word underneath (see
Stimulus items were also selected in order to investigate anecdotal reports about variation and change in their usage. Observation suggests that signs for country names, for example, have been undergoing considerable change in recent years. For UK placenames, anecdotal reports from BSL teachers indicated that lexical variants differ according to in-group and out-group membership.
For each target concept, participants were asked to produce the sign variant they use most on a daily basis. One limitation of this task is that sign variants are elicited in isolation and signers may respond to normative pressures or perceptions about what is appropriate for their region rather than offer the sign they actually use most. Therefore, a subset of the conversational data was also analysed to investigate whether the sign variants from the lexical elicitation task correspond to those used in the free conversation data (collected prior to the lexical elicitation task).
Lexical variants for colours, countries and numbers (33 concepts) were elicited from 249 participants, producing a total of 8, 217 tokens. All tokens were annotated in ELAN
Phonological variants were grouped together as one lexeme
The lexical elicitation task produced an extremely complex dataset in which each of the individual stimuli represent a variable that could be investigated in detail. As a means to capturing overall patterns in the data, we coded each response as either a ‘traditional’ or ‘non-traditional sign’ for the signer's region. This would make it possible to investigate whether there is any evidence that levelling is taking place in BSL, and whether or not this language change is more strongly associated with any specific subgroup(s) (e.g., young males) in the British deaf community. The traditional signs for colours, countries and numbers for each region were determined by two methods. First, there are existing teaching resources about BSL lexical variation that make claims about the association of certain BSL signs with particular regions in the UK. For example, the lexical variant
Each token was coded for the following social factors: age, gender and social class of the signer. For this study, participants were categorised into three age groups: 16–39 years (younger), 40–59 years (middle) and 60 years and above (older) to reflect the different educational policies experienced by deaf children throughout the twentieth century (and thus our age groups reflect emic criteria
In addition to these social factors, the semantic category of the elicited sign (signs for countries, numbers or colours) was coded to see whether subcategories of signs were changing at a different rate than others.
In summary, sign variants produced for colours, countries and numbers were coded as traditional or non-traditional for the signer's region. The following social factors were investigated: region (Belfast, Birmingham, Bristol, Cardiff, Glasgow, London, Manchester, Newcastle), age (16–39, 40–59, 60+), gender (female, male), social class (middle, working), language background (deaf, hearing), school location (local, non-local). The semantic category of the sign was also coded (colour, country, number).
The UK placename data were analysed separately to investigate anecdotal claims about their usage. Such claims suggest that place name signs may work to index local, in-group versus non-local, out-group identity. For example, it is claimed that Bristol signers use a different lexical variant for ‘Bristol’ than those living elsewhere. To investigate this claim, fieldworkers from each of the eight regions were asked to identify which lexical variants were considered to be the local variants. The elicited forms for these UK placenames were analysed and the variants produced were coded as either local or non-local for the particular placename. In most cases, the local sign is a lexicalised form of fingerspelling consisting of the first letter and possibly some subsequent letters (e.g., the Manchester variant meaning ‘Manchester’ is formed by producing the manual letter ‘M’ followed by ‘C’) (for more discussion of lexicalised fingerspelling, see
The effects of the observer's paradox were likely to be greater during the lexical elicitation task than the conversational task, due to the relatively greater attention to sign language production in the former compared to the latter activity
For the current study, we carried out multivariate statistical analyses of the data using Rbrul
The results for the colour, country and number signs study are discussed first. This study investigates the relationship between the use of traditional regional signs and social factors (e.g., signers' age, gender, etc.). Following this, the results for the UK placenames study will be presented. Finally, the comparison between the conversational and lexical elicitation data will be discussed.
Of the 6722 tokens analysed, 5279 (79%) were classified as traditional for the signer's region. Participant and lexical item were included in the analysis as random effects.
Factor Group | Factor | Log odds | Tokens | % of traditional signs | Centred weight |
16–39 | −0.969 | 2074 | 66.1 | 0.275 | |
Non-local | −0.281 | 2924 | 75.0 | 0.430 | |
Hearing | −0.223 | 4623 | 78.7 | 0.444 | |
Semantic category | |||||
Countries | −0.426 | 1623 | 72.6 | 0.395 | |
Social class | |||||
Working | −0.09 | 4123 | 79.1 | 0.478 | |
Gender | |||||
Female | −0.018 | 3501 | 79.0 | 0.495 |
Application value: Traditional signs.
*Factor groups significant at p<.05. 6722 tokens.
Input probability = 0.866, Mean = 0.785, Intercept = 1.868, Deviance = 5752.805. Random (participant) standard deviation = 1.064. Random (lexical item) standard deviation = 0.809.
We tested for interactions between the seven variables under investigation and found that region and school location were not independent of each other (region/school, p>0.05); as a result, we excluded region of residence from the analysis. Of the six factor groups remaining, the following three, in order from greater to lesser importance, predict the use of traditional signs: age, school location and language background. Participants in the older age group strongly favour the use of traditional signs (0.693), while those in the younger age group strongly disfavour the use of traditional signs (0.275). Those who were educated locally slightly favour the use of traditional signs (0.57) compared to those who were educated outside of the region where they reside (0.43). The third most significant predictor was language background, with participants with hearing parents slightly disfavouring the use of traditional signs (0.444). Participants with deaf parents slightly favour the use of traditional signs (0.556). The semantic category of the sign, social class and gender were not found to be significant. The results for each category (in increasing order of their proportion of traditional signs: signs for countries, numbers and colours) were analysed separately to look at the patterns of traditional sign use.
In the country names dataset, a total of 1623 tokens were analysed.
Factor Group | Factor | Log odds | Tokens | % of traditional signs | Centred weight |
16–39 | −0.832 | 501 | 58.9 | 0.303 | |
Language Background | |||||
Hearing | −0.212 | 1108 | 72.2 | 0.499 | |
School location | |||||
Non-local | −0.165 | 707 | 70.4 | 0.459 | |
Social class | |||||
Middle | −0.005 | 620 | 71.1 | 0.499 | |
Gender | |||||
Male | −0.04 | 780 | 71.1 | 0.49 |
Application value: Traditional signs.
*Factor groups significant at p<.05. 1623 tokens.
Input probability = 0.793, Mean = 0.726, Intercept = −1.342, Deviance = 1646.468. Random effects (participant) standard deviation = 0.809. Random effects (lexical item) standard deviation = 1.131.
Age, school location and language background were significant predictors of the use of traditional number signs (see
Factor Group | Factor | Log odds | Tokens | % of traditional signs | Centred weight |
16–39 years | −1.316 | 1195 | 66.3 | 0.211 | |
Non-local | −0.408 | 1682 | 75.2 | 0.399 | |
Hearing | −0.33 | 2683 | 79.6 | 0.418 | |
Social class | |||||
Working | −0.169 | 2371 | 80.0 | 0.458 | |
Gender | |||||
Female | −0.1 | 2028 | 79.6 | 0.475 |
Application value: Traditional signs.
*Factor groups significant at p<.05. 3877 tokens.
Input probability = 0.909, Mean = 0.794, Intercept = 2.301, Deviance = 3018.346. Random (participant) standard deviation = 1.654. Random (lexical item) standard deviation = 0.697.
Of the 1222 colour tokens analysed, only 16% (201 tokens) were non-traditional sign variants. The signer's age and school location were important in predicting the use of traditional colour signs (see
Factor Group | Factor | Log odds | Tokens | % of traditional signs | Centred weight |
16–39 | −0.692 | 378 | 75.1 | 0.334 | |
Non-local | −0.224 | 535 | 80.4 | 0.444 | |
Gender | |||||
Female | −0.045 | 632 | 83.7 | 0.489 | |
Language background | |||||
Hearing | −0.035 | 832 | 84.4 | 0.491 | |
Social class | |||||
Working | −0.032 | 749 | 84.0 | 0.492 |
Application value: Traditional signs.
*Factor groups significant at p<.05. 1222 tokens.
Input probability = 0.885, Mean = 0.836, Intercept = 2.037, Deviance = 990.857. Random effects (participant) standard deviation = 0.898. Random effects (lexical item) standard deviation = 0.682.
To investigate anecdotal reports that individuals who reside within a given urban centre use a different sign variant to refer to that city than non-residents do, a total of 1992 tokens were classified as either local or non-local for the particular placename analysed. The results revealed that, with the exception of ‘Glasgow’, ‘London’ and ‘Manchester’, the use of the local placename variant significantly correlated with residency in that location, with residents found to strongly favour the use of the local variant: Belfast (>0.999), Birmingham (0.673), Bristol (0.736), Cardiff (0.724) and Newcastle (0.794).
Lexical item | Factor | Log odds | Tokens | Centred Weight |
Belfast | ||||
Non-residents | −8.395 | 219 | <0.001 | |
Birmingham | ||||
Non-residents | −0.72 | 219 | 0.327 | |
Bristol | ||||
Non-residents | −1.024 | 217 | 0.264 | |
Cardiff | ||||
Non-residents | −1.017 | 217 | 0.266 | |
Newcastle | ||||
Non-residents | −1.347 | 219 | 0.206 |
Application value: Local variant for region. All factor groups significant at p<.05.
This study has investigated the lexical variants produced by 249 signers for 41 concepts within the semantic fields of colours, numerals, countries, and UK placenames. In total, 10, 209 tokens were elicited. Of these, 295 separate lexical variants were identified for the 41 stimuli. Thus, the variation for these concepts in BSL is considerable, mirroring the findings of previous studies
Number of stimuli | ||||||||
Number of variants | 0 | 1 | 2 | 3 | 4 | 5 | 6 | 7 |
1 | x | |||||||
2 | x | |||||||
3 | x | |||||||
4 | x | |||||||
5 | x | |||||||
6 | x | |||||||
7 | x | |||||||
8 | x | |||||||
9 | x | |||||||
10 | x | |||||||
11 | x | |||||||
12 | x | |||||||
13+ | x | |||||||
TOTAL |
To consider the use of traditional signs across settings all examples of the signs for colours, countries and numbers from the LFS conversational data were compared to those elicited as part of the lexical elicitation task. A total of 570 tokens were coded for analysis. Of these, 124 tokens (22%) were not the same sign variant as those elicited during the lexical elicitation task, suggesting that 78% of elicitations as part of the lexical elicitation task were an accurate representation of the signer's actual lexical use when there is less attention paid to their language production. There were 26 instances in which a response involved a signer using a traditional variant in one setting and another traditional variant in a different setting, or a non-traditional variant in one setting and a different non-traditional variant in another setting. Twelve percent of responses (15 of 124) involved the use of a non-traditional variant in the lexical elicitation task and a traditional variant in the conversational task. In sixty-seven percent of responses (83 of 124) a signer used a traditional variant in the lexical elicitation task and a non-traditional variant in the conversational task. Overall, the results indicate that the majority of signers used the same variant across different settings (conversation and the lexical elicitation tasks). In conversation, the more naturalistic setting, a minority of participants produced more non-traditional variants.
Participants' use of traditional signs was conditioned by a number of social factors, in order of significance: age, school location, and language background. In this section, we discuss these findings in relation to previous sign language research, explore whether the results reveal language change in progress, and finally suggest what this study can tell us about the BSL variation in the future.
The results suggest that age is the most significant factor predicting the use of traditional signs with a decline in use from older to younger signers. This was true, to differing degrees, across all regions. Some variants were found to be unique to certain age groups. For example, some number sign variants which originate from Irish Sign Language were not present at all in the data elicited from younger participants. These variants are associated with St. Vincent's School for deaf children in Glasgow, a Catholic school in which teachers used Irish Sign Language as the language of instruction until the 1950s
School location is the second most important factor predicting the use of traditional signs in BSL. Those individuals who were educated locally in our study used a higher proportion of regional signs than those individuals who had attended a school outside the region. The findings do suggest that, although individuals may have lived in a given region for the last 10 years, they do not entirely adapt their lexicon to the local variety. This finding appears to confirm the importance of schooling in the maintenance of regional variation in BSL, supporting Quinn's
Language background is also a significant factor in predicting the use of non-traditional signs. Signers with deaf parents favour the use of traditional signs, supporting previous work on ASL which found that signers with deaf parents favour the use of ‘conservative’ variants
Although semantic category was not a significant factor in predicting the use of traditional signs, the results indicate that some semantic categories are undergoing greater change than others. In this study, signs for countries are changing at a faster rate than signs for numbers and colours.
Signs for countries appear to have undergone the most dramatic change. Around half of the responses by younger participants in the lexical elicitation task (302 tokens) were not traditional signs for their region. Younger and older signers may have adopted a different lexical variant later in life from the variant acquired in their early years. This is quite evident from the discussions between participants during the lexical elicitation task in which many older and younger signers explicitly mention the sign they formerly used and the sign they use now.
By comparing the traditional variant for each region to the most frequent non-traditional variant used amongst younger signers, we can observe the direction of change in some country name signs. For example, the traditional BSL sign meaning ‘China’ is produced at the eyes with a twist of the wrist of both hands (see
The change in progress for signs meaning ‘China’ appears to be the same across all eight regions in the BSL Corpus Project. This shift in signs representing China co-occurred with the introduction of signs for other east Asian countries. The pattern of change (from a sign representing physical characteristics to one representing clothing) is not observed for other country name signs. For example, the BSL variant
Additionally, a small number of signers in the younger and middle groups are using a sign variant borrowed from Indian Sign Language
Lexical borrowing may be the source of several of the country signs in this study as we have seen with ‘India’. Like ‘India’, the traditional variants for ‘America’ in all regions have been replaced amongst younger signers with the borrowed ASL form
Finally, not all borrowed forms may come from the corresponding country's sign language. For example, we find that seven regions (again, except for Belfast which already uses
To summarise, signs for countries in BSL have experienced rapid lexical innovation. This may, in some cases, be linked to pressures from political correctness but it may also reflect increased contact between deaf communities around the world leading to the borrowing of signs. The language change processes for signs for numbers and colours will be discussed in the next section.
Previous research has suggested that signs acquired first are generally maintained over a signer's lifetime
In the dataset presented here, we have analysed the changes in colour signs as well as number signs. Results must be viewed with caution as the colour dataset is significantly smaller than the number dataset (17 target numeral stimuli compared to 5 target colour stimuli), but the results indicate that there have been fewer changes in colour signs than number signs. Age was found to be a contributing factor in the use of colour signs with younger signers using a decreasing proportion of traditional signs compared to older signers. Instead, those younger signers who used non-traditional signs used variants associated with the London region (e.g.,
With UK placenames, we were interested in the folk belief that residents of a city use a different sign variant for their city's name than non-residents. The only exceptions were the signs for ‘Glasgow’, ‘London’ and ‘Manchester’: almost all signers in all regions used the same lexical variant for ‘Glasgow’ (221 tokens), ‘London’ (244 tokens) and ‘Manchester’ (209 tokens). In most cases, the endonym, or local name for the city, was a reduced fingerspelled form, as described above, with the exception of the sign for ‘Belfast’. The exonym, or name used for the city by individuals from outside that city, in some cases was a calque – i.e. a literal translation of the equivalent English words. For example, one exonym variant for ‘Manchester’ consists of a compound of individual signs
In this investigation of lexical variation and change in signs for 41 key concepts in BSL, we have identified a number of processes taking place in the language that reflect different conditioning factors operating on different subsets of the data. Age is an important predictor of lexical variation and change across all groups of signs in this study. Signs for countries are subject to a number of external influences, including political correctness, changing attitudes towards lexical borrowing, and greater international mobility and transnational contact. The change we see here is age-graded, with anecdotal evidence of some older signers also adopting newly introduced variants. Changes in the use of traditional regional signs for colours and numbers, however, do not appear to be subject to changes in attitudes to language, but appear to reflect changes in the transmission of BSL as well as increased mobility within the UK and exposure to lexical variation in BSL via the media.
This descriptive and quantitative analysis of a large dataset of BSL provides a ‘snapshot’ of BSL lexical variation and change synchronically and suggests how societal changes have directly influenced BSL. This study lays the groundwork for more detailed ethnographic studies and investigations into the relationship between variation, change and language attitudes in the British deaf community.
We gratefully acknowledge the following people: the project co-investigators (Margaret Deuchar, Frances Elton, Donall O'Baoill, Rachel Sutton-Spence, Graham Turner); Daniel Ezra Johnson for his continued support with the data analysis using RBrul; the Deaf Community Advisory Group members for their advice and assistance (Lorna Allsop, Linda Day, Clark Denmark, Helen Foulkes, Melinda Napier, Tessa Padden, Gary Quinn and Kate Rowley); the project's research associate (Sally Reynolds) and to our deaf community fieldworkers who made the data collection possible (Jeff Brattan-Wilson, Mischa Cooke, Avril Hepner, Sarah Lawrence, Dawn Marshall, Evelyn McFarland, Carolyn Nabarro, Melinda Napier, Mark Nelson, Jacqueline Parker, and Jenny Wilkins). We are grateful to Jackie Carcillo for cross-checking the data coding for this study and to John Holliday for his continued support in using ArcGIS mapping. Finally, we extend special thanks to all of the deaf individuals who participated in the BSL Corpus Project: without their participation, none of this research would have been possible.