SLA Topics Vivian Cook
Perceptions of Class and Age in English
[Sorry Tables have gone wrong in html]
The issue to be looked at here is how English people associate certain words with speakers of different social classes and ages1. Starting with class, the relationship between class and language has in recent years been discussed mostly in terms of phonology rather than grammar or vocabulary, for example the work with Norwich accents (Trudgill, 1974). Yet a speaker who says Pardon? rather than Sorry?, who answers How do you do? with Fine or Pleased to meet you, or who talks of serviettes rather than table napkins was, according to my upbringing in Southern England, infallibly giving away their working class status. Liza Doolittle’s use of Garn and Not bloody likely! was as much a sign of her Cockney origins as her accent. The popular controversy over U (Upper-class) and Non-U (non-Upper-class) initiated by Alan Ross in the 1950s (Ross, 1956) reflected an everyday aspect of life to British people in the 1950s—the perception of class—which some claim is now less prominent. The main questions here are therefore the extent to which English people in the 1990s still categorise other people in class terms through their use of vocabulary and which words are associated with particular classes.
The links between age and language have also primarily been seen through phonology. For example people react to a sentence read by an older-sounding voice with a standard accent as ‘egocentric, living in the past and talking of trivia’, read by a young voice as ‘arrogant and pompous’, and read by an elderly voice with a non-standard accent as ‘stupid, and losing his grip’ (Giles & Coupland, 1991). Most English speakers become aware that some aspects of their vocabulary give away their age, say the use of wireless rather than radio, gay meaning ‘merry’, or grotty for disapproval. The main questions are again the extent to which English people categorise other people’s age through their vocabulary and which words are associated with which ages.
The emphasis here is on class and age as perceived by the speakers. It is not so much whether people of a particular socio-economic class or age-group actually say some word but whether it is perceived by speakers as a marker of class or age. As Cameron (1995) argues, people’s beliefs about language are as crucial as the ‘objective’ facts of their usage; much of the gender debate has concerned what people ascribe to men or to women, not what they actually say. Meyerhoff (1993) too calls for research into how people ‘perceive particular terms to be out-group markers’. It is nevertheless fairly conservative to accept speakers’ judgements only as evidence for how they perceive English rather than what they know about English when much linguistics research is predicated on speakers’ judgements (Birdsong, 1989).
Classes are treated here as labels that people apply to each other rather than precise socio-economic terms. The contrast is between MC (middle class) and WC (working class), reflecting the everyday distinction. Age is divided into five bands: 1-17, 18-24, 25-49, 50-64, 65+, again to provide divisions that speakers can handle, contrasting people of school age at one end with senior citizens at the other and dividing the intervening years approximately into student, young and middle-aged.
Choice of vocabulary items
The choice of vocabulary items to test was based
on areas that were believed to show a difference between classes and generations
in British English, some having shown up in earlier pilot research (Cook, 1997),
such as terms of approval and disapproval. Some of the differences were
anticipated to be arbitrary lexical variation between two items, one
associated with younger age or lower class, the other with older age or higher
class. Other differences were anticipated to reflect certain plausible
1) The use of “y” in names for family relations would be more middle class than the short forms, for example daddy versus dad.
2) First names often have a difference between the full form, Deborah, the “y/ie” form, Debbie, and a short form Debs. The patterning of this is complex and depends on specific names with in general the “y/ie” forms suggesting the person is being regarded as more childish (Wierzbicka, 1992). Hence the short and “y/ie” forms should be associated with younger people rather than the full forms. The jazz player of the 1950s Johnny Dankworth became the musician of the 1980s John Dankworth. Additionally spellings in “i” would be associated with young people, i.e. Susi rather than Susie (though not to the extent of testing the double “i” form favoured by Danii Minogue).
3) words for approval/disapproval are likely indicators of class and age since they may proclaim membership of a particular group—say excellent versus fab.
4) items dealing with new technology should be susceptible to age differences if only because much has been introduced during the lifetimes of the older people; that is to say, LP should go with older people, CD with younger.
5) full forms of words or names would be associated with older, more middle-class, people, for example refrigerator versus fridge, television versus telly, bicycle versus bike/cycle
Other word sets were added that were believed to be associated with class or age, such as words for ‘male person’, cinema, drinks, ‘bicycle’ and drunkenness. Ross (1956, p.25) claimed indeed that ‘cycle is non-U, against U bike, bicycle’ .
The final selection of 68 items, based on British usage, was:
family relations (8): mummy, mum, daddy, dad, granny, gran, nanny,
first name variations with ‘short’ and ‘ie/y’ forms, some only
in spelling (14): Robert, Robbo, Robbie,
Susan, Sue, Susie, Susi, John, Johnno, Johnny, Deborah, Debbi, Debbie, Debs
iii. ‘approval/disapproval’ (11): all right, smashing, great, horrible, grotty, bad, good, not bad, terrible, fab, excellent
‘refrigerator’ (2): fridge, refrigerator;
‘radio/television’ (8): television, TV, telly, tube, box, tranny, wireless,
radio; ‘record’ (8): disc, CD, LP,
vinyl, record, record-player, deck, stereo
other areas. ‘male person’ (5): chap,
guy, mush, bloke, lad; ‘drunk’
(3): sloshed, pissed, drunk; ‘tea’
(3): cup of tea, cup of char, cuppa;
‘bicycle’ (3): bike, cycle, bicycle;
‘cinema’ (3): movies, flicks,
The method was a questionnaire distributed through organisations in the Colchester area of Essex, on the one hand schools or colleges dealing with the young, that is to say either Sixth Forms of Colchester schools or the Colchester Sixth Form College, or with the old, that is to say the University of the Third Age at the University of Essex. The primary division was then into age-groups of young and old speakers; some regrouping into class groups was also possible, as will be seen below, through this was not part of the main design. This is similar to the methodology used by Chambers (1995) in dialect topography in that, while it is ‘messy’ in using an arbitrary set of people and in recording their beliefs rather than their actual usage in depth, it permits relatively quick access to a reasonable sample size. It is also similar to the testing of lexical pairs in Bayard (1989) and Meyerhoff (1993) except that the latter used oral questions rather than a written questionnaire. One limitation of a questionnaire is that it tests variation in the written form rather than the spoken, for example missing some alternative forms for mother such as /m¾m/ or /m‰m/. One minor advantage is that it permits the testing of alternative spellings, for example Debbie versus Debbi.
The questionnaire had two main sections. In part A the people had to circle the class and age-band for the 68 words, the instruction being In the following questions, you have to guess the class (Middle Class (MC)/Working Class (WC)) and the age-group of the speakers of the words in bold type. This was deliberately an either/or forced choice between MC and WC. Each test word was given in a context sentence, as in the example Hello, auntie. Each context sentence was in a normal conversational style rather than being either very formal or informal and they were told Imagine you can hear people speaking to other people they know well—a friend or member of their family.
Part B consisted of 19 multiple choice questions; people had to tick which of the following words you would use yourself when speaking to a friend or a member of your family, for example What’s the time (a) dad, (b) daddy, (c) other ......... The answers had at least three alternatives, one always being ‘other .....’, with an instruction If none of them are right, give the word you would use instead. Part B used a selection of the words from Part A but its open alternative permitted a range of other words, partly as a check on whether the words chosen in Part A were representative of people’s usage. Within the two parts, the questions were in randomised order.
The distribution yielded 72 completed questionnaires from people aged 13-17 and 42 from those aged over 50, plus 16 in the 18-49 age-groups, not considered here; these will be referred to from now on as the Young and Old groups respectively. All the people were native speakers of English living in Colchester, a town in the county of Essex some 50 miles north-east of London. Of the Young group 52 claimed to have been born in Essex and of the Old group 16; about half of the others put down the neighbouring areas of Suffolk, Norfolk, London or East Anglia in general. In terms of self-ascribed class, the Young group assigned themselves 38.9% to the WC, 55.6% to the MC; the Old group 23.8% to the WC, 80.9% to the MC. While the Old group indeed had mostly middle-class occupations such as teaching, the Young group were all students. Because they were contacted through post-compulsory education, both groups were comparatively well-educated. Given that Colchester has had a large influx of older population and that young people tend to leave the town after secondary school, this is probably as balanced a pair of age-groups as could be obtained in this locality.
Most of the respondents seemed to take the task seriously. The Young group may have occasionally been tempted to exaggerate: it is hard to believe that any teenager actually calls their mother woman as one reported; the diverse range of expressions they gave for drunkenness may reflect a desire to shock rather than true linguistic diversity. The only word that caused any problem was mush, which seemed to be absent from Essex speech.
The scoring of the questionnaires for perceived
class was straightforward. A small proportion of people, 6.8%, did not indicate
a single class for any given item, mostly because they circled both classes. A
chi-squared test was carried out on each word for each group to establish
significant differences between scores (goodness of fit, df.1). Table A shows
the words that were class-marked for one class or the other by both groups,
together with the words that were unmarked. The majority of words, namely 43/68,
were perceived in the same way by both groups, 27 MC, 15 WC, with 13 being
Table B displays the words that differed between
the groups. A further chi-squared test was performed on the combined figures for
each word for both groups (df.1).
Only eight differences involved a shift of class: dad, fridge, stereo and Robbie were thought WC by the Young group and MC by the Old; smashing, deck, CD and TV were MC for the Young group and WC for the Old. The remaining two words were unmarked for one group, marked for the other: radio is MC for the Old, Unmarked for the Young; mush is WC for the Old, Unmarked for the Young.
The scoring for perceived age was also straightforward, totals being calculated for each age-band; answers that ringed more than one age were excluded, bringing the total of non-answers to 9.8%. The chi-squared test was used as a test of goodness of fit to establish significant differences (df.4). treating each word for each group separately. All the results for both groups were significant at p<.005 except for p<.025 cycle (O, Y), bike (O), TV (O), box (O), great (O), terrible (O), record (O), Deborah (O), and non-significant flicks (O), telly (O), horrible (O), bad (O), vinyl (O), deck (O), Johnny (O), tube (Y). In other words, both groups perceived the words as being related to particular age-groups, with only seven words out of 68 that were unmarked for the Old group, one for the Young.
One unexpected problem concerned words for family
relations. Virtually all these words were perceived as spoken by Young people by
both groups; for instance 65 out of 72 of the Young group assigned daddy to band 1-17, 35 out of 42 of the
Old group. Clearly these words are associated, not with age-related word choice,
but with real world constraints on people of given ages. A person who has a
mother or a grandmother is perceived as young; most people over 65
presumably do not need to use these terms as they would not have surviving
relatives. Usage is linked to the facts of human life, so to speak. The only
exception seems to be dad for the Old
group, which was distributed fairly evenly over the first three age bands
1‑49. Words for relations were then
excluded from the discussion of age here and not included in Table C.
The next step is to look at words that are overwhelmingly related to one of the five age-bands by taking a score of 50% of the answers indicating a given age-band. Using this criterion, the Young group perceived as under 18 (band 1) all right, sloshed and pissed; as 18-24 (2) bloke, CD, Robbo, Robbie, Debbi, Debbie, Debs; as 25-49 (3) terrible, Susan, Sue; as 50-64 (4) chap; and as 65+ (5) wireless. Using this criterion the Old group only located one word at an age-band, namely 18-24 (2) Debbi.
Taking the score for a single age-band as the criterion ignores the spread over the other bands. For example the following graphs illustrate the patterns for wireless, all right and bloke, all statistically significant at p<0.005, here expressed in percentages to even out the totals for the two groups.
Here the pattern is clearly skewed towards the right; wireless is seen as the property of the older age-bands 4 and 5 by everyone.
The pattern for all right is, however, clearly skewed towards the left; all right is seen as the property of the younger age-bands 1 and 2 by everyone.
The pattern for bloke on the other hand reveals a difference between the two groups, skewed to the left for the Young group, peaking at 2, but having a smoother bell-shaped pattern for the Older group, peaking at 3. Hence bloke seems younger for the Young group, and middle-aged for the Old group.
To summarise the skewing in these patterns without displaying graphs for each word, the five bands were collapsed into three, with some loss of information: Young (bands 1+2), Middle-aged (band 3), Old (bands 4-5), locating a word in whichever of these three had the highest total. Table C displays those words where the perceptions of both groups were in agreement. Only those words are displayed that were statistically significant, as explained above.
Some words are then perceived by both groups as primarily Young, such as mush and bike, some expressions of attitude, such as fab, and some ‘short’ forms of first names such as Robbo and Debs.; others such as chap and LP are perceived as Old; three as Middle-aged. 20 out of the 68 words were then allocated by both groups to a distinct age.
Table D shows the words where there was disagreement between the perceptions of the two groups. This was again tested by carrying out a chi-squared test on the combined figures for each word for both groups (df.4).
The differences are mostly where a word perceived as Old by the Old group is seen by the Young group as Young (bloke) or as Middle-Aged (bicycle). In the first case, users simply see their own group as using a word and perhaps do not realise that it is used by another group. Only one word tranny was seen as Young by the Old group and Old by the Young group, perhaps showing that both groups regard it as used by someone other than themselves, i.e. it is not used much by anybody any more. Thus, although many words were allocated to age-bands, the precise age differed between the groups for 21 out of the 68 words.
Comparison with words supplied
Let us now turn to the words that the people reported saying themselves. Again the totals for each question for each group were tested with the chi-squared test and were significant at p<0.01 for each of the 19 questions (goodness of fit test, df. from 2 to 4 depending on the question).
The words which had a major difference between Young and Old are shown in Table E. These are grouped according to how great the difference was in percentages rounded to the nearest whole number.
It is evident that 15 words differ to a considerable extent according to the age of the group. The interest lies in whether these agree with the perceptions of age, in other words whether these correspond either to the unanimous judgement of age seen in Table C or to the divided perceptions seen in Table D. The common judgments in table C include chap (seen as Old) and LP (seen as Old). The divided judgements in D include: bloke, drunk and television. It is evident that the most frequently produced forms were not the same as those perceived as belonging to one age.
Talking only in terms of the most frequent response, however, conceals the frequency of other responses; a word may still be associated with a particular age-group even if it is not produced the most often. One form of evidence is the words that are said with zero frequency by one group but fairly often by the other: these include for the Old group tipsy (40%), record-player (55%), chap (45%), and wireless (14%), for the Young group vinyl (11%) and not bad (10%). Perhaps these are better age markers. Chap and wireless are certainly ascribed to the Old, as seen in table C, but the perception of the other words was either divided (not bad) or unmarked (record-player, vinyl).
Since the people were also asked their class, it is possible to link their productions with their self-adjudged class, even if the groups were not primarily chosen for that reason. Regrouping the subjects led to two groups of 69 Middle Class and 33 Working Class. Using the chi-squared test to compare scores for each group per question, only one comparison was significant (df.4, p<0.005), namely words for ‘drunk’ where pissed was proportionately much higher for the WC, compared with drunk and tipsy for the MC as seen in Fig 4, adjusted into percentages.
The other procedure used above was to search for words
used by one group to some extent but never by the other. On this criterion daddy occurred 7 times for MC, 0 for WC,
sloshed and cycle 5 time from MC, none by WC.
However, as the WC group was much smaller than the MC group, this may reflect
the effects of individual variation. Overall then both classes actually produced
much the same words.
The overall question about the extent to which English people perceive class-marking in vocabulary is readily answered: they have firm associations between certain words and particular classes. Words such as mum and bike stand out as Working class for both groups, words such as drunk and cinema as Middle class. Rightly or wrongly, English people still associate certain words with certain classes.
The question of the extent to which they perceive age-marking in vocabulary is also answered positively: English people have clear beliefs that certain words such as wireless and chap go with older people and other words such as all right and pissed go with younger people. This perception does not necessarily accord with their own reported usage.
We can now turn to the specific hypotheses outlined at the beginning:
1) were “y” names for family relations perceived
as class-marked? The “y” forms mummy,
daddy and granny were seen as MC,
the reduced forms mum, and nan as WC; gran and nanny were unmarked and dad differed between the groups (WC to
the Young, MC to the Old). With the exception of nanny, to be discussed below, the “y”
form of relatives is clearly marked as MC. As confirmation, the card on Princess
Diana’s coffin was addressed to Mummy.
2) were full forms of first names related to the middle class and the old? The full forms Robert, John, Deborah, Susan were all perceived as MC by both groups. The short forms Robbo, Debs, Johnno were seen as WC by both groups but Sue was MC for both groups. Divided usage was found for Robbie (WC by Young, MC by Old); the other “i/ie” forms Debbi, Debbie, Susie were unmarked. MC speakers are then perceived as using more full first names, WC speakers more short forms. The “i/ie” ending did not distinguish between the classes. In terms of age, however, the short forms Robbo, Johnno and Debs were seen as Young by both groups, as were the “i/ie” forms Debbi, Debbie and Susi, apart from Susie which was divided. So there are indeed tendencies towards the full forms for MC and the “i/ie” forms for Old.
3) were words for approval/disapproval age and class-related? In class terms excellent, great, horrible, terrible, and good were seen as MC by both groups, all right as WC; smashing differed from MC for the Young to WC for the Old. In age terms Young words for both groups were all right, fab, grotty; divided words were good, excellent, not bad (Young for the Young, Old for the Old) and horrible, terrible (Young for the Young group, Middle-Aged for the Old group). The only addition from the production test was the Old group’s use of crap. Several words of approval/disapproval are strongly related to class or age, sometimes both—all right apparently stereotypes the speaker as young and working class. The production data included a range of occasional forms not in part A, mostly from the Young, terms of approval such as cool, brill, dreadful, fantastic, brilliant, blinding, really good, pretty fucking good, OK and terms of disapproval such as awful, bad, rubbish, shit, wank, pathetic, and poo.
4) were words for technology age-related? Wireless and LP were clearly seen as Old by both groups; they differed over stereo, television, TV and tranny. The pattern does not demonstrate that the newest technology goes with Young, the oldest with Old, apart perhaps from wireless and LP; however, vinyl (the older type of record) was only supplied by the Young group but in this meaning only dates from 1976 according to the OED.
5) were full forms of words associated more with older middle-class speakers? In class terms both groups indeed classified refrigerator, bicycle, television and record-player as MC and bike, telly and tranny as WC, although fridge was divided. In age both groups found bike Young and cycle Old, and were divided over refrigerator and television. The word bike then suggests an older middle-class speaker, despite the claim by Ross (1956). Otherwise the tendency seems to be for full forms to be seen as more MC. This is supported by the perception of full first names as MC in (1).
As we have seen, while there is a strong belief that a given word is MC or WC, Young or Old, this varies to a great extent from one word to another. This section brings together some information on particular words that have not been mentioned earlier.
One issue may be the extent to which words for relations vary dialectally. Take words for ‘grandmother’. Though granny is in general the most wide-spread word in England (Upton, Sanderson, and Widdowson, 1987), granma is shown for the Essex area; nanny and variants nan and nanna are shown in North Norfolk, some 80 miles due north of Colchester. Figure 5 shows the pattern for the people here , the comparison of the scores for both groups being significant (chi-squared test, p<0.005, df. 5). The Old group indeed prefer granny (55%), gran (23%) and grandma (26%), with nan (7%) and nanna not occurring. But the most popular word for the Young group was nan (42%), followed by nanna (17%) and nanny (3%), with low frequency for granny (4%), gran (17%), and grandma (13%). In other words 59% used a form that supposedly occurs at the geographically opposite end of East Anglia. In addition two people used the unusual forms granny myn and nin.
Looking more closely at the subgroup born in
Essex, 56% used nanna/nan/nanny, that
is to say there is no appreciable difference between native Essex and non-native
Essex. Either the restriction of nanny forms to North Norfolk reported in
the Survey of English Dialects is wrong or this area of vocabulary has changed
for the Young group since the days when the Survey of English Dialects data were
collected. Indeed the later volume the Survey of English Dialects: the Dictionary
and Grammar (Upton, Parry & Widdowson, 199X) records /n¾n/
from Essex. In terms of class perception both groups mark granny as MC, nan as WC; in age terms, no difference
emerged because of the perception that only younger speakers talk about older
relations. An interesting point that emerged from discussions with some
local people is that they claim to use one term for their maternal
grandmother, one for their paternal grandmother, which they believe
unique to their family, similar to the Swedish distinction between forfar and morfar. Clearly these terms need further
investigation in other ways.
The most entertaining part of the production data
was the words for ‘drunk’ which included a large range of forms such as, from
the Old, completely sozzled, and,
from the Young, slaughtered, paralytic,
legless, fucked, wrecked, plastered, hammered, nutted, out of ‘er ‘ead, drunk as
a skunk, and out of it.
cinema itself is MC, flicks WC. Movies is not perceived as marked for age or class and occurs only 3 times in the production data. In New Zealand Meyerhoff (1993) found a switch from pictures to movies, which she regards as US English, not apparently influential in England.
This section discusses some general aspects of change in vocabulary. A common model of linguistic change balances the upward thrust of status with the downward thrust of solidarity with the working class (Trudgill, 1974). So, in class terms, the perception of chap as middle-class could reflect a prestige form—come on, chaps; bloke a downward pull to solidarity—a good bloke. This opposing-tendency model makes no clear predictions, as a word could have gravitated downwards or upwards since its introduction. The concept of lexical diffusion might seem appropriate where sound changes percolate through changes in individual words (Chambers & Trudgill, 1980), but it is difficult to apply this model logically to lexical items without clear evidence of consistent movement from class to class. The Labovian notion that new words are introduced in informal styles and then spread to less formal styles would explain the apparent shift in bike from U to WC over forty years.
Turning to age, the question is where change comes from. One possibility is that words are arbitrarily associated with age-groups and that they go forward with the age-group cohort as they get older: wireless came in, say, as a young person’s term and is now only found in the old as they have retained their original usage. The logical consequence of this, however, is that many words should die with their speakers; genuinely old forms must in principle disappear. Yet, as we see below, most age-related terms have existed in the language for many years. Chap as a general word for person is first found in the OED for 1716, even wireless is 1894. In both cases the first generation to use them are long-buried yet they are nevertheless associated with the old.
A second alternative is the concept of age-grading in language first described by Labov (1964): speakers adopt a form of language suitable to their years, just as in television advertisements retired men put on cardigans to potter in the garden. So chap may be permanently associated with the old and is adopted when, say, the speaker passes 50. Young people may say something because they believe it is a sign of their youth, for example pissed and telly. As Labov (1994, p.73) says, ‘adolescents and young adults use stigmatised variants more freely than middle-aged speakers, especially when they are being observed’, particularly true here of words for ‘drunk’ and for disapproval. This can explain the discrepancies between the words perceived and the words supplied: people feel that sloshed is a a badge of the young, cycle a badge for the old, even if they do not report using them themselves.
Some pertinent informn for both class and age may be the dates of first occurrence of words recorded in the Oxford English Dictionary (1989). Often the rival words are equally old: bicycle, thought MC by both groups, is dated 1868, cycle, thought MC, is found first in 1870, and bike, thought MC, comes from 1882. MC cinema, itself an abbreviation of cinematograph, first occurred in 1899, WC flicks in 1926, and movies in 1912; MC Old wireless is 1894, MC radio 1903. MC stereo and WC tranny are 1964 and 1969 respectively. smashing and fab are both unmarked though coming from 1911 and 1961 respectively. This group of words hardly demonstrates that the older words are distinctively associated currently with class or age. In a second group of words the oldest word is indeed more likely to be MC or Old. The MC Old word chap is recorded from 1716 in a rather ‘low’ meaning; WC bloke is 1851; mush is 1936 and unmarked guy in the sense of ‘man’ is 1848 (and found in D.H. Lawrence in 1928). The ‘inebriated’ sense of MC drunk goes back to 1340, the WC pissed to 1929 and sloshed to 1946. Overall then there is no clear chronological pattern to suggest that the present-day Old and MC words came from earlier forms than the present-day Young and WC words.
To sum up, this research attempted to see how people categorise each other through the vocabulary they use. Its use of vocabulary is novel in that most previous studies of age and class have concerned phonological or grammatical differences. Its use of perceived class may be contentious, even if such perceptions are widely used as evidence in other areas of language. Its choice of vocabulary items may be disputable, covering a diversity of items rather than a precise set, even if it shows that indeed much of the variation is particular to a given word. Nevertheless it shows something of the potential interest in exploring people’s associations of words with class and age.
1. I am endebted for comments and statistical help to David Britain, Denise Chapell and Phil Scholfield and for their participation to the staff and students of the Philip Morant School, the Greyfriars Institute and the Colchester Sixth Form College, all in Colchester
Bayard, D. (1989), ‘“Me say that? No way”: the social correlates of American lexical diffusion in New Zealand English’, Te Reo, 32, 17-60
Birdsong, D. 1989. Metalinguistic Performance and Interlanguage Competence. New York: Springer.
Cameron, D. (1995), Verbal Hygiene, London; Routledge
Chambers, J.K. (1995), ‘An introduction to dialect topography’, English World-Wide, 15, 1, 35-53
Chambers, J.K. & Trudgill, P. (1980), Dialectology, Cambridge University Press
Cook, V.J. (1997), Inside Language, London; Arnold
Giles, H., and Coupland, N. (1991), Language: Contexts and Consequences, Open University Press; Milton Keynes
Labov, W. (1964), ‘Stages in the acquisition of standard English’, in R. Shuy (ed.), Social Dialects and Language Learning, Champaign, Illinois; NCTE
Labov, W. (1994), Principles of Linguistic Change. Volume 1: Internal Factors, Oxford; Blackwell
Meyerhoff, M. (1993), ‘Lexical shift in working class New Zealand English: variations in the use of lexical pairs’, English World-Wide, 14, 2, 231‑248
Oxford English Dictionary (1989), second edition edited by Simpson & Weiner, Oxford University Press
Ross, A.S.C. (1956), ‘U and Non-U’, in Mitford, U. (ed.) Noblesse Oblige, Hamish Hamilton. London, reprinted Penguin, Harmondsworth (1959), 9-32
Trudgill, P. (1974), The Social Differentiation of English in Norwich, CUP
Upton, C., Parry, D. & Widdowson, J. (199X), The Survey of English Dialects: the Dictionary and Grammar, XXXXX
Upton, C., Sanderson, S. & Widdowson, J. (1987), Wordmaps, Croom Helm, London
Wierzbicka, A. (1992), Semantics, Culture, and Cognition, Oxford University Press, Oxford