The relation between the ageing process and language use has traditionally been analysed
from two points of view: the changing language used during the lifespan of an individual,
and the language of different cohorts of individuals living within a speech community.
Age-specific use of language refers to the first approach, with generation-specific use of
language reserved for the latter. The distinction has been an important one in the analysis
of language change.
Women and men
cannot be categorized in the same way in this society, however, since for women the
relevant age-related social categories involve marital status, with widows who do not
remarry assuming an important role that is otherwise restricted to men (ibid.). This point
is valid generally, because in all societies gender and other relevant social variables
interact with the age variable, however it is defined, making comparisons between
different age groups far from straightforward.
Age may be a more meaningful social category in some cultures than in others. Ota,
Harwood, Williams and Takai (2000, 34) found that for 18-19 year olds a group identity
in terms of age was stronger for Americans than for Japanese. Furthermore, being young.
Eckert (ibid) points out that “only the middle aged are seen as engaging in mature use, as “doing” language rather than learning it or
losing it”.
More commonly, however, the term refers to age specific differences in a more general sense, characterising the language consideredappropriate to and typical of different stages in the life span.
Age-grading may also involve the use of age-preferential features which are used by
speakers of all ages in the community, but more frequently by some age groups than
others. Age-grading is not necessarily associated with language change, since individuals
may change their language during their lifetime, whilst the community as a whole does
not change (Labov 1994, 84). Indeed, one of the greatest challenges in the analysis of
language change in progress is deciding when age differences in language use reflect a
change in community norms, and when they reflect stable age grading (see section 3)
This pattern of age differentiation is assumed to be the normal pattern for stable
sociolinguistic variables, although it cannot be assumed that it will be found universally,
especially, Chambers and Trudgill say (op. cit., 79) if social conditions are different. A
particularly striking recurrent observation is that adolescent speakers from all social
classes in a wide range of urban communities use a significantly higher number of
variants that are socially stigmatized than do speakers of other ages. Figure 3, from
Wolfram and Fasold’s early Detroit study (1974, 91) illustrates this pattern for the use of
multiple negation (where the stigmatized form would be she don’t want nothing, and the
prestige variant she doesn’t want anything). Wolfram and Fasold comment that a similar
distribution could be indicated for “any number of phonological or grammatical features”
(1974, 91). Similar findings have been reported for other languages (see, for example,
Silva-Corvalan (1981) for Spanish).
The study of generation-specific language acquires a special significance for research
into language maintenance, language death, code-switching and borrowing. Language
shift typically takes place over three generations. For example, a migrant group in
London from, say, Turkey, may arrive in London with Turkish as their first language and
will begin to acquire English as a second language. Their children will then be bilingual
in Turkish and English, using each of the languages in different domains (English, for
example, at school and Turkish with their parents) and may also codeswitch between
Turkish and English. Their children may then speak only English. This, of course, is an
idealised model but it indicates the kinds of important differences that may be found
between the different generations in this type of situation.
Research into the language of different generations of bilingual speakers can contribute to important theoretical questions. Budzhak-Jones and Poplack (1997), for example, use data from two generations of Ukrainian –English bilingual speakers to address the vexed question of whether single word items from one language occurring in a stretch of discourse from another language should be considered loan words or single word codeswitches. The first generation speakers in their small study were aged between 57 and 76, had been living in Canada for at least forty years, and used Ukrainian as their primary language in the course of ‘normal daily interaction’ (op.cit., 230). The second- generation speakers were aged between 20 and 31, and used Ukrainian sporadically, mostly in communication with older people. For them English was their primary language of interaction, although like the first generation speakers they said that they had acquired Ukrainian as their first language. Using the stepwise multiple regression procedure incorporated in Goldvarb 2.0 (Rand and Sankoff 1990), Budzhak-Jones and Poplack analysed 399 English-origin nominals and a comparative sample of 481 Ukrainian nominals. Their analysis suggested that the first generation speakers fully controlled the quantitative conditioning of linguistic variation in the marking of nominal inflection, but that the second generation speakers did not. Budzhak-Jones and Poplack were therefore able to argue that it is essential to distinguish ‘native’ speakers from ‘non- native’ speakers in the study of bilingual discourse. The first generation speakers in their sample, they claim, could be considered native speakers whereas the second generation speakers could not. Items borrowed from English, even if only for the nonce, were fully integrated by the first generation speakers into Ukrainian, and should therefore be considered integrated loanwords rather than single-word codeswitches.
A different example of generation-specific language comes from the research of Dubois and Horvath (1999) on language change in Cajun English. Dubois and Horvath analysed the use of seven phonological variables in the speech of three generations of speakers from rural areas of Louisiana. During the period of history represented by the three generations language shift from French to English has been taking place. The sociolinguistic situation, Dubois and Horvath say, is one of a stable closed community in the process of language shift, such as are found in many urban areas of the world, including, for example, the barrios of Los Angeles and New York City, and the migrant enclaves in Sydney. Dubois and Horvath’s findings force them to question several of the assumptions of variationist research on sound change, such as the principle that women tend to be innovators in change from below, as well as leaders in change from above. Dubois and Horvath show that in Cajun English the distribution of sound changes across the generations is intimately tied to the external events that have shaped the lives of the three generations of speakers. For example, English was made the compulsory language of education during the period when the older generation would have been of school age; for them, therefore, English is a second language, and any social meanings attached to variation are realized in French (op.cit., 304). Their English has little to do with the usual understanding of change except for the important fact that their ways of speaking provided the source for future changes within the community. The middle-aged generation experienced the effects of industrialization and urbanization and the consequent imposition of an external norm for the speaking of English. The younger
generation, on the other hand, lived through a period of Cajun renaissance where a Cajun identity became salient but could be signalled only through English.
Living out a life against the backdrop of these different historical events defined different social and economic roles for men and women at different historical times, so that the effect of the basic social variables of social network and gender on language variation was different for each generation. The inclusion of age as a variable, then, was profoundly important for understanding the effects of these familiar social variables. As Dubois and Horvath point out, age as a measure of an individual’s chronological development was not what was important; instead age “is fundamentally important to identify the generations within the speech community affected by important historical events. The effects of gender are strongly conditioned by generation, and the generations are strongly conditioned by sociohistorical contexts” (op.cit., 311).
4. Social implications
The most obvious social problems concerning age and generation-specific differences in language are connected with education. One such problem resulting from age-grading behaviour could affect children from families that do not expect them to take part in adult conversations (such as the working class Black community referred to as ‘Trackton” by Heath (1983) or those British families where children are expected to speak only when spoken to). Clearly these children will be less forthcoming in oral discussions in the classroom than children from families that have treated as more equal partners in conversation. For discussion of these and other cultural differences, and their educational implications, see Romaine (1984, 159-228).
The pattern of preferential age grading whereby adolescents use a higher proportion of stigmatized variants than speakers of other ages may also impinge on school performance. Teachers and parents may react negatively if they perceive a heavy use of stigmatized features and do not realise that this is likely to be a temporary phase. The association between the use of nonstandard linguistic features and stereotypes of laziness,