Age Grading in Sociolinguistic Theory

Distinguishing linguistic change at the community level (‘generational change’) from linguistic change at the individual level (‘age grading’) is ‘‘one of the major issues in contemporary sociolinguistics’’ (Tagliamonte 2012:247). This article gives a brief history of the study of language change in the community, before turning to the types of linguistic behavior that have been observed across individuals’ lifespans. The article also discusses the meanings that have been attributed to the term ‘age grading’, arguing that consensus cannot be reached without more longitudinal work to determine the limits of lifespan linguistic change. 1. The Study of Language Change The goal of linguists who study language change is to understand its mechanisms and motivations. To do this, they must first reliably identify that a language change has actually occurred or is still occurring. This article will first briefly describe the techniques developed by linguists to capture historical and ongoing language change. It will then explain how, in the latter case, change in progress might be misdiagnosed. Rather than having observed a linguistic innovation increasing in a community over time (‘generational change’), the linguist might have captured a regular association of a variant with a particular lifestage (‘age grading’). A third possibility, which is perhaps the most common, is that both community and individual change are occurring simultaneously. This article will give examples of all three patterns, and discuss why they are important to a general understanding of language change. Human languages arise through a combination of universal shared capacities (Chomsky 1957) and the social interactions of individuals and communities. Like any other aspect of community social behavior, language is as subject to change over time as clothing, music, government policies, gender norms etc. The search for the mechanisms of language change was until the early 20th century the occupation of historical linguists. By carefully examining the writings of earlier generations or civilizations, historical linguists were able, through a process called ‘comparative reconstruction’, to make educated guesses at how, for instance, spoken Latin evolved into its daughter languages such as French, Spanish, Portuguese and Romanian, or how the vowels of English underwent a dramatic reorganization in the late Middle Ages. This painstaking work has showed us that many diachronic changes (that is, operating over time) have progressed with a great deal of linguistic regularity. The identification of conditioning environments (e.g. following nasal consonants, following high front vowels) that favor certain kinds of historical changes helped to pave the way for the rule-based approach of synchronic (that is, relating to a particular point in time) generativist linguistics in the mid-20th century. Often missing from historical accounts of language change, however, was an understanding of the social factors that motivate language change. Post-hoc descriptions of Language and Linguistics Compass 6/6 (2012): 371–382, 10.1002/lnc3.343 a 2012 The Author Language and Linguistics Compass a 2012 Blackwell Publishing Ltd chains of historical events (e.g. [bon] > [bõn] > [bõ] for the nasalization of French vowels) with explanations based solely on linguistic environment (e.g. nasalization occurred because of following nasal consonants, followed by deletion of the consonant) did not provide a good account of why these changes might have occurred in the first place. This shortcoming of the comparative reconstruction approach was highlighted by Uriel Weinreich, William Labov and Marvin Herzog (henceforth WLH) in their 1968 publication Empirical Foundations for a Theory of Language Change (Weinreich et al. 1968). They identified five key problems that required engagement through a different empirical method. The first is the ‘constraints’ problem: What kinds of linguistic changes are possible and which are not? Historical evidence can only show us which changes have already occurred, but not which changes are impossible. The second is the ‘embedding’ problem: How does a language change reach everybody (or almost everybody) in a given community? How is it embedded, not only in linguistic environments, but in social environments? Historical texts often represent only a very tiny slice of the population who spoke a language, and thus cannot adequately help us to understand the spread of language changes. The third is the ‘evaluation’ problem: How do community members react to language changes, and what effect to their evaluations have on the progress or nature of a change? Social evaluation is often recoverable from the writings of commentators on language, but tends to be anecdotal and the commentators are usually drawn from a restricted social stratum. The fourth is the ‘transition’ problem: How does a language change from one state at time A to another state at time B? Finally, the fifth problem is ‘actuation’: why does a particular language change take place when and where it does, and not at some other time or place? WLH argued that these important problems could only be addressed by observing language change as it happened, and from multiple social vantage points, including a wide range of ages, social classes and localities. Scholars working in the framework of WLH are known today as ‘sociolinguists’ or ‘variationists’. 2. The Apparent Time Construct It was William Labov who pioneered a methodology for observing language change that he has summarized as ‘the use of the present to explain the past’ (Labov 1978). Labov showed that one could examine the use of a changing language feature in a cross-section of a speech community by using a wide range of ages as a proxy for historical time (e.g. Labov 1966 [2006], 2001). In order to interpret the distribution of the language feature, the researcher relies on the assumption that individuals’ speech patterns are largely fixed by early adulthood. Older people’s use of the language feature represents the typical use of that feature in the community when they themselves were young. Use of the language feature by pre-adult speakers represents the use of the feature in the community at the time of the study. In New York City in the 1960s, for example, Labov (1966 [2006]) looked at the frequency of use of post-vocalic ⁄ r ⁄ in e.g. car, guard. Older members of the community exhibited lower overall rates of use of ⁄ r ⁄ than did younger members of the community. Labov interpreted the differential rates of ⁄ r ⁄ use across ages as evidence for a rise in ⁄ r ⁄ use over time, as the community moved away from the local pattern of r-lessness toward the majority American pattern of r-fulness. His inference was supported by historical evidence from dialect atlases and from the comments of various observers which showed that New York City had earlier exhibited much higher rates of r-lessness than it did in the 1960s. In keeping with the program of research defined in WLH, Labov also demonstrated that it was possible to draw conclusions about the social motivations 372 Suzanne Evans Wagner a 2012 The Author Language and Linguistics Compass 6/6 (2012): 371–382, 10.1002/lnc3.343 Language and Linguistics Compass a 2012 Blackwell Publishing Ltd for language change if one had sufficient information about speakers’ social class backgrounds, gender, ethnicity, and their attitudes to local speech. This groundbreaking work opened the door for many scores of similar studies of other communities across the world, all confirming that it was possible to capture and examine ongoing language change, and that one could advance some explanations for its actuation, embedding, constraints and evaluation (for an overview, see e.g. Meyerhoff 2011). Sociolinguistic insights and methods have also been usefully employed by historical linguists working on textual materials (e.g. Kroch et al. 2000; Schneider 2003), such that the boundaries between historical linguistics and sociolinguistics remain porous. The interpretation of what has come to be known as the ‘apparent time construct’ relies centrally on the assumption mentioned earlier that individuals stabilize their linguistic repertoires after adolescence. Work on the critical period for language learning (e.g. Lenneberg 1967) supported the idea that many aspects of the language system are immutable after a certain age. As we will see, numerous sociolinguistic studies have demonstrated the continuing utility of the apparent time construct (Bailey et al. 1991), even as some have also shown that individuals do not always ‘fix’ their speech patterns in adulthood. The main question to be asked in this essay is how we can identify and categorize types of individual linguistic instability in order to better understand how it affects our interpretation of apparent time. 3. Generational Change or Age Grading? Individual Community Synchronic Pattern 1 Stability Stable Stable flat 2 Age-grading Unstable Stable monotonic slope with age 3 Generational change Stable Unstable monotonic slope with age 4 Communal change Unstable Unstable flat 5 Lifespan change Unstable Unstable monotonic slope with age Patterns of change in the individual and the community. Adapted from Sankoff (2005:1004, table 103.1). What kinds of individual linguistic change occur over time, and how can we recognize them? The table above represents Labov’s (1994:83) summary of the space of logical possiblities, with one additional contribution from Sankoff (2005) in italics. Case (1), stability, comprises a community that is unchanging and homogenous with respect to the linguistic feature under study. The ‘default’ case of interest to sociolinguists is (3), generational change. The definition of generational change given here assumes that individuals are linguistically stable over their adult lifetimes, and language change increments over time in the community through the addition of new generations who adopt quantitatively stronger manifestations of an innovation e.g. ever-increasing frequencies of ⁄ r ⁄ in NYC. In two cases out of the first four, individuals’ speech patterns are ‘unstable’ (i.e. changing): (2) age grading and (4) communal change. In the definition presented by this table, (2) age grading is the instability of an individual’s use of a feature over the lifespan against a backdrop of community stability for the same feature. Well-known examples of variables that exhibit this distribution in apparent time are (ing) and multiple negation (Labov 2001; Trudgill 1974; Wolfram and Fasold 1974), and they are supported by diachronic evidence that the alternation between variants has existed in the Age Grading in Sociolinguistic Theory 373 a 2012 The Author Language and Linguistics Compass 6/6 (2012): 371–382, 10.1002/lnc3.343 Language and Linguistics Compass a 2012 Blackwell Publishing Ltd communities for a very long time, with neither appearing to replace the other (Houston 1985; Nevalainen 1998). According to Labov, (4) ‘communal change’ occurs when all individuals (i.e. the entire community) change their use of a linguistic feature concurrently, yielding a flat distribution across age groups. Labov’s examples include the rapid – almost simultaneous – uptake of a lexical item by all community members (Payne 1976). Yet Labov also mentions a case that did not affect all individuals simultaneously: Arnaud’s (1998) study of the innovative English progressive – ing in the 19th century showed that some individuals increased their use of this form along with the community, while others remained stable. Since the publication of Labov’s table, other sociolinguists have narrowed the definition of ‘communal change’ to just the rapid, community-wide cases, with e.g. Meyerhoff (2011) giving the example of community-wide, simultaneous adoption of a taboo on the name of a deceased person. Meanwhile, Arnaud-type cases of slow community generational change coupled with more rapid individual change in the same direction were labeled by sociolinguists (including, incidentally, Labov) as a special type of age grading, as in (1). Because this collapsed cases of community stability with community change, Gillian Sankoff proposed adding a fifth line to the table, for the term (5) ‘lifespan change’. She defines this as cases in which both individuals and community are unstable over time, and where individuals ‘‘change over their lifespans in the direction of a change in progress in the rest of the community’’ (Sankoff 2005:1011). ‘Age grading’ is then reserved for cases in which individuals are unstable and the community is stable, as per Labov’s original definition. As we will see in later sections, this division is not completely satisfactory. One methodological problem is that ‘age grading’ is indistinguishable from (3) generational change in progress when only apparent time evidence is available. As we saw, Labov’s study of ⁄ r ⁄ in New York City revealed increasing use of ⁄ r ⁄ with decreasing age. At first blush, this suggested either that young people simply preferred to use ⁄ r ⁄ , abandoning it along with in-group slang as they grew older (age grading), or that young and old people represented the contemporary and historical states of the community respectively (generational change). Only by comparing the contemporary linguistic situation with evidence from the past could Labov assert that a generational change was taking place. Of course, since he had no information about the speech of individual New Yorkers as they aged, Labov could not go on to distinguish generational change (3) from generational change accompanied by individual change (lifespan change, 5). Yet linguists are not always so lucky regarding diachronic community evidence. Sometimes only fragmentary historical evidence is available, and the analyst must rely on judicious interpretation of apparent time alone. The distribution of a variant across social classes and genders can be revealing (Cameron 2005). Since women have been repeatedly identified as the leaders of generational language change (Labov 1990), apparent time studies in which men are the most frequent users of a variant might be more indicative of age grading than change. Non-standard variants of (ing) and (neg), for example, often occur at higher rates in the speech of men than women in every age group (Labov 1966 [2006]; Trudgill 1972). The differential behavior of social classes across age groups can also be diagnostic of age grading rather than apparent time. In a generational change, upper social classes have been implicated in the vanguard of prestigious ‘changes from above’ while lower social classes lead unnoticed ‘change from below’ (Labov 1990). In all cases, however, the leading social groups are followed by the others, as the linguistic change diffuses through the social hierarchy, and this can be seen in parallel trend lines. 374 Suzanne Evans Wagner a 2012 The Author Language and Linguistics Compass 6/6 (2012): 371–382, 10.1002/lnc3.343 Language and Linguistics Compass a 2012 Blackwell Publishing Ltd When one or more social classes appear to be stable, or moving in a different direction to the rest, however, then age grading is a possible interpretation. In one well-known example, Macaulay’s (1977) study of stigmatized glottal stop in Glasgow English, males in the highest and lowest social classes exhibited stability in apparent time, while the second-highest social class showed a peak in early adolescence in use of glottal stop, followed by rapid decline in adulthood. This pattern of behavior in apparent time was independently interpreted as age-grading by Sankoff (2004) and Chambers (2003). Further support for this interpretation is provided by the behavior of the girls and women in the sample, who (unlike the boys) exhibited a uniform decrease in glottal stop use with age in the higher social classes, without the peak in adolescence exhibited by the boys. Both Sankoff and Chambers proposed that the peak represents stronger pressure on middle class boys than on girls to orient to covert, non-standard norms of behavior in the teenage years. They did so because they shared assumptions about the social pressures experienced by men and women from different social classes at different stages of the life course. Central to these assumptions is the concept of ‘linguistic market’. 4. The Linguistic Market across the Life Course The ‘linguistic market’ is an alternative definition of social class. Its originators, Bourdieu and Boltanski (1975) argued that existing macro-definitions of socioeconomic class derived from income level, occupation, educational background, etc. did not distinguish speakers within the same class whose access to and use of the standard language in a community were quite different. Chambers (2003) illustrates this by pointing out that among an occupational group labeled ‘professionals’ one might find both those for whom adherence to standard linguistic norms is less important (chemists, engineers) and those for whom that adherence is highly important (teachers, lawyers). Sankoff and Laberge (1978) developed a measure of placement within the linguistic marketplace, whereby raters were given descriptions of the 120 speakers in the Sankoff-Cedergren Montreal corpus (Sankoff and Sankoff 1973) and asked to rank them according to the extent to which the standard dialect of the community was important in their lives. The speaker descriptions included not only details of their occupational history, but also that of their spouses and parents where applicable. This line of research has been important in helping us to understand the differential pressures to use standard language that exist across status groups, as well as acknowledging that this pressure may increase and decrease across a speaker’s lifespan. Apparent time studies of stable variables such as (ing) have shown a curvilinear pattern with age, with a peak in the use of non-standard forms in the teenage years (e.g. Labov 2001). Adolescents have been characterized as relatively free of responsibilities and normative pressures from the linguistic market, and this is assumed to be reflected in their high rates of use of nonstandard features. Middle-aged adults, on the other hand, have usually undergone what Chambers (2003:195) calls linguistic ‘‘retrenchment’’: a retreat from the non-standard variants used in youth followed by stabilization. Speakers in this life stage have greater responsibilities at work and at home, and have slowed down their earlier frenetic attempts to ‘define’ themselves, becoming relatively settled in their tastes and opinions. Interpreting apparent time age distributions through the lens of the linguistic market is intuitively very appealing, but requires additional sophistication if it is to be fully convincing. Eckert (1997:159) remarks that sociolinguists have come to see language development over the lifespan as principally motivated by individuals’ increasing awareness of the standard language, but other factors are involved. Individuals’ speech may also reflect their Age Grading in Sociolinguistic Theory 375 a 2012 The Author Language and Linguistics Compass 6/6 (2012): 371–382, 10.1002/lnc3.343 Language and Linguistics Compass a 2012 Blackwell Publishing Ltd developing awareness of other salient local vernacular dialects, or of the gender norms of their community, or of aspects of their own social identity. A perspective on lifespan linguistic change that considers only the individual’s (fluid) position in the standard talk market does not capture the full picture. In addition, Eckert argues, we do not yet know enough about the linguistic generalizability of the life stages to which we commonly make reference, such as childhood, adolescence, middle age and old age. More work is needed, at finer granularity, on all of these stages and the transitions between them. Eckert herself has provided a wealth of detail on the linguistic lives of adolescents and pre-adolescents (Eckert 2000, 2009) that she has supported with ethnographic observations on the general and specific social pressures that shape their language use. A similar approach has been taken by Rose (2008), who combined quantitative analysis of linguistic variation and change with ethnographic fieldwork in a senior center in Wisconsin. Work on children’s acquisition of socially meaningful variation has also advanced (e.g. Renn 2011; Roberts 1994; Smith et al. 2007). The middle years, however, remain in Eckert’s (1997:165) words, something of a ‘‘vast wasteland in the study of variation’’: a default, ‘normative’ life-stage against which other life-stages are compared. She appeals for new approaches: We need to divide up the middle years to see them as developmental and graded too. How is anticipating a promotion from manager to vice president different from anticipating the leap from elementary to junior high? (Eckert 1997:158) We might also ask about the transition from full-time work to retirement, the transition from stay-at-home parent to the workplace, and countless others. A further concern is that the life course unfolds against the passing of historical time: the social meaning of ‘elderly’ or ‘retired’ in the mid-twentieth century, for instance, may be different to the meaning of those categories (and the behavioral expectations associated with them) for the retiring Baby Boomer generation at the turn of the 21st century. Two further problems with linguistic market explanations for individuals’ linguistic adjustments across the lifespan are that (i) without detailed ethnographic observations they are vulnerable to post-hoc generalizations and that (ii) very few such adjustments have been clearly supported with real time evidence. Only by following a panel of individuals over some portion of the lifecourse can we (i) firmly conclude that linguistic adjustments have occurred and (ii) refine our interpretations of those adjustments. Chambers (2003:199) suggests that the optimal study –at least of ‘retrenchment’ – would involve tracking a large number of 15-year-olds into their early thirties, and recording and observing them in a variety of contexts (e.g. school, work, home). The emphasis on multiple contexts is important, since individuals’ stylistic repertoires can be very complex (Sharma and Rampton 2011) and must interact with their average rate of use of a variant across their lifespans. No-one has yet met Chambers’ scholarly challenge, but in the next section, we will review some studies in which the mechanisms and motivations for lifespan change and age grading have been explored in real time. 5. New Evidence from Trend and Panel Studies Longitudinal studies can take two forms: panel studies, in which the same individuals are recorded at multiple points, and trend studies, in which a sample of different but comparable individuals are recorded at each point in time. As the field of sociolinguistics has matured, it has become possible for researchers to return to earlier field sites to carry out trend studies. The increasing availability of digitized archives of text and speech from non-sociolinguistic sources has also made 376 Suzanne Evans Wagner a 2012 The Author Language and Linguistics Compass 6/6 (2012): 371–382, 10.1002/lnc3.343 Language and Linguistics Compass a 2012 Blackwell Publishing Ltd longitudinal studies a realistic prospect. Among the trend studies carried out by sociolinguists, a few are notable for having indicated that lifespan linguistic change could potentially occur alongside community generational change. Pope and colleagues’ re-study of Martha’s Vineyard (Labov 1963; Pope et al. 2007), inferred individual lifespan change by comparing the linguistic behavior of each age cohort at two time points. An overall increase in the raising of the nucleus of (aw) on Martha’s Vineyard, for instance, was shown to have been accompanied by an increase in every comparable cohort. Speakers born in 1932–1948, for example, exhibited some raising in 1962, but had increased the level of raising by 2002. Pope and colleagues attribute the increased raising over 40 years not only to incrementation with each new generation, but also to individual linguistic adjustments in the direction of community change as adults aged. Yet one can only infer lifespan change from trend studies. Panel studies are needed to confirm it. In recent years, a steady trickle of panel studies has accumulated, although some date from as early as the 1980s (e.g. Prince 1987). Because of the inherent difficulty involved in relocating and reinterviewing large numbers of subjects, the majority of panel studies have been restricted, typically comprising one subject (Carter 2007; Harrington et al. 2000; Rickford and McNair-Knox 1994; Tagliamonte 2008) or a small handful of subjects (Baugh 1996; De Decker 2006). Larger scale studies have been principally limited to studies of Montreal French (Thibault and Vincent 1990), to studies of Brazilian Portuguese (e.g. Scherre and Naro 2006), the LANCHART project in Denmark (Gregersen 2009), historical text studies (e.g. Nevalainen et al. 2011) and studies that ‘piggyback’ on large-scale longitudinal projects in other academic fields (e.g. Van Hofwegen and Wolfram 2010). What are the implications of panel study results for our understanding of age grading? Firstly, panel studies suggest that in the majority of generational community changes in progress, some individuals make post-adolescent adjustments to their speech in the direction of community change (Sankoff and Blondeau 2007; De Decker 2006; Harrington et al. 2000; Scherre and Naro 2011; Wagner 2008). Secondly, the vast majority of individuals remain stable after adolescence, providing support for the apparent time construct. Thirdly, some individuals exhibit retrograde behavior during a community change, favoring conservative forms as they age and retreating from innovative ones. Prichard and Tamminga’s (2011) apparent time study of young adult Philadelphians, combined with Labov’s (2011) trend study, suggests that individuals who attend nationally oriented institutions of higher education are the most likely to reverse their speech away from the direction of local sound changes: a hypothesis also tested in Wagner’s (2008) panel study of young adult Philadelphians. Wagner and Sankoff (2011) used a panel study of Montreal French speakers to demonstrate that some individuals decrease their use of the innovative periphrastic future tense form as they age, in favor of the conservative inflected future form. This brings us again to the problem of ontology and terminology. Apparent time and real time evidence have already shown us that community change and individual lifespan change can be found in a number of different configurations. Since change at both of these levels appears to intersect in many cases, does it still make sense to provide separate labels for ‘generational change’ at the community level and ‘lifespan change’ at the individual level? Furthermore, should we continue to differentiate the possible types of individual change as ‘lifespan change’ and ‘age grading’? I address these questions in the next section. Age Grading in Sociolinguistic Theory 377 a 2012 The Author Language and Linguistics Compass 6/6 (2012): 371–382, 10.1002/lnc3.343 Language and Linguistics Compass a 2012 Blackwell Publishing Ltd 6. Toward a Definition of Age Grading Since the adoption of ‘age grading’ from anthropology by Hockett (1950:453), sociolinguists have refined the term several times, but inconsistency in its application remains. 6.1. AGE GRADING IN INDIVIDUALS VERSUS A STABLE COMMUNITY Labov (1994:83) defined age grading as individual linguistic change against a backdrop of community stability. This definition has the advantage of objective applicability: if the community can be shown conclusively to be stable with respect to use of a variable, then any individual change over the lifespan can be classified as age-grading. Unfortunately, very few cases of community linguistic stability have been confirmed, besides the aforementioned (ing) which has been stable for centuries (Houston 1985) and multiple negation, which has long been a stable minority means of negating English clauses (Nevalainen 1998). One notable exception is José’s (2010) trend confirmation that devoicing of ⁄ z ⁄ is a stable variable in Indiana. Future temporal reference in Montreal French may also be stable after centuries of change from inflected to periphrastic forms, with the periphrastic firmly established as the default variant, but further longitudinal evidence is required to confirm this (Wagner and Sankoff 2011). 6.2. AGE GRADING AS REPETITION OF A PATTERN Sankoff (2005) acknowledged that age grading could be applied to situations beyond those of community stability. She described it as ‘‘those situations in which groups of speakers in the same community, generation after generation, find it appropriate to employ a particular pattern .... Such patterns may often be associated with stable variables’’ (Sankoff 2005:1011). Under this definition, age grading is identified as a repetition of age-appropriate linguistic behavior in each generation whether or not the linguistic variable is stable in the community. Cheshire (2006:1553) also defines age grading as a repetitive pattern: ‘‘a change of behavior with age that repeats itself in every generation’’. Sankoff explicitly contrasts this with her term ‘lifespan change’, which she applies to those cases in which individuals change along with the community. Yet in cases of slow linguistic change, individuals might change along with the community in every generation. How many times must this happen for the pattern to be described as ‘age grading’ rather than ‘lifespan change’? Furthermore, the best way to determine whether an ageassociated pattern repeats over the generations is to collect longitudinal data. Yet almost all of the studies mentioned heretofore collected samples from communities or individuals only twice. A pattern repeated at two intervals might not be evident at a third interval (Poplack and Lealess 2011), either because the pattern occurred by coincidence, or because the state of the community has changed, e.g. from stable to changing, or from one direction of change to a reversal. Thus the ‘repeating pattern’ definition of age grading seems insufficient at our present state of knowledge, due to the methodological difficulties associated with attesting a pattern. 6.3. AGE GRADING AND MARKETPLACE PRESSURE As we have seen, age graded variables have been associated with community stability, and repeating patterns. A third, overlapping criterion for identifying age grading is the social value of the sociolinguistic variable in the linguistic market. As we have seen, this is 378 Suzanne Evans Wagner a 2012 The Author Language and Linguistics Compass 6/6 (2012): 371–382, 10.1002/lnc3.343 Language and Linguistics Compass a 2012 Blackwell Publishing Ltd typically conceived of as the extent to which a variant is overtly evaluated as ‘standard’ within a community, with a curvilinear distribution of the non-standard variant across age groups. For this to happen, the variable is likely to have been operative in the community for a long period of time, and thus to be stable or close to stable, as is the case for stigmatized tense variants of (aeh) in Philadelphia, which Labov (2001) refers to as a ‘nearly-completed change’. However, the notion of linguistic market is contested and problematized, as we have seen (Eckert 1997). Our methods for determining the social salience and social value of linguistic variables have been considerably improved through recent advances in experimental techniques (e.g. Campbell-Kibler 2005). But categorizing a putative change as ‘age-graded’ rather than ‘lifespan change’ would presumably require us to draw a line between those variables that are especially sensitive to marketplace pressure and those that are not. How and where would such a line be drawn? Complicating this criterion still further is the current paucity of panel studies of individuals’ participation in unremarked-upon ‘change from below’.