«Journal of Sociolinguistics 4/3, 2000: 323±347 Styling the worker: Gender and the commodi®cation of language in the globalized service economy1 ...»
a normative construct which shapes, even if it cannot wholly determine, the behaviour of those language-users to whom it is prescribed.4 THE DATA My data were collected for a larger project (Cameron 2000) which looked at some range of service workplaces, but for the purposes of this article I concentrate on a single type of workplace, namely the `call centre', an institution in which people are employed to make or take telephone calls. (The `make or take' distinction is captured in the industry terms `outbound' and `inbound' to refer to call centres where employees either initiate or receive calls. The centres I looked at were exclusively `inbound', i.e. calls were initiated by customers. Outbound call centres typically have sales rather than service as their prime function, whereas I was most interested in the provision of customer services.) I chose to study call centres, in particular, for two reasons. First, they provide a prototypical example of a `new' service workplace: the vast majority have existed for less than ten years, and their institutional culture has always incorporated the disciplines of globalized capitalism. Call centres as we now know them came into existence when it was recognized that advances in telephony and computing enabled customer service functions traditionally performed locally (e.g. in each branch of a bank or travel agency) to be # Blackwell Publishers Ltd.
The work of call centre operators is notoriously stressful, being both extremely repetitive and subject to demanding performance targets, and this is re¯ected in high rates of employee turnover in the industry (Carter 1998; Reardon 1996).
Media coverage of call centres has been both copious and generally critical, often suggesting that they are the sweatshops of the 21st century (Wazir 1999).
Second, language has a special signi®cance in call centre work. The operator's job consists of little else but language-using ± talking to customers on the phone and inputting/retrieving data using a computer ± and her/his professional persona must be created entirely through speech. Typically, the speech of call centre operators is subject to intensive regulation and constant surveillance. Supervisors can covertly listen in on any call (known in the industry as `silent listening'), while in some centres every call is recorded and may become the subject of `counselling' (a worker and a supervisor or manager listen together to examples of the worker's performance and engage in critical assessment).5 Call centres, then, are a good example of service work as language work, and as such they are also a particularly rich source of insight into the commodi®cation and regulation of language on the job.
I collected data relating to seven centres located in various parts of the U.K.
(central Scotland, the north of England and London). The service functions performed in these centres were: providing directory assistance to telephone subscribers, logging faults in telecommunications equipment, dealing with auto insurance claims, processing personal banking transactions, authorizing credit requests, booking rail tickets and handling enquiries for a utility (gas) company.
The data at my disposal take the form of notes on observations, tapes/transcripts of interviews, and copies of written materials including employee manuals, training packs, appraisal forms and lists of criteria for assessing performance, scripts and prompt sheets for standard work routines, and memos discussing linguistic issues.
The analysis in this article draws most heavily on the last-mentioned of these data-types, namely the textual materials. These provide the clearest and most detailed picture of what linguistic ideal a call centre `ocially' wants its operators to aim for, what it prescribes and what it proscribes. However, it should be noted here that I was able to obtain a suitable quantity and quality of texts from only four of the seven centres in my sample; these four therefore # Blackwell Publishers Ltd. 2000 STYLING THE WORKER 329 dominate the analysis presented below (it will be obvious which they are from my attribution of extracts). Of the other three centres, one did not as yet produce detailed written speci®cations for linguistic performance; in the remaining two cases I was not able to procure copies of the speci®cations from the sources available to me (see further below).
Some supplementary information comes from a set of interviews conducted between May and December 1998 (the main purpose of interviewing was to elicit insiders' perceptions of call centre work; since that is not the focus of the present article, the use made of interview data here is limited). I interviewed four call centre managers, two supervisors and six operators, employed in ®ve dierent centres located in central Scotland, northern England and London (these are a subset of the seven mentioned above). All interviews were conducted individually, in most cases face to face but in two cases on the phone. (All but two took place `o-site', for reasons explained below.) Interviews were `semi-structured' ± I had a schedule of questions, but I encouraged informants to respond at length where they had more to say, and to introduce additional concerns. Each interview lasted at least 30 minutes.
It will be evident from the details just given that dierent centres I had dealings with provided dierent kinds and quantities of information. This re¯ects some problems associated with researching commercial enterprises in general and call centres in particular.6 In cases where I undertook observation in a call centre I did so with the co-operation of the management, but there were often restrictive conditions attached. Because of the critical media coverage I have already mentioned, I found many managers concerned about negative publicity, which led some to want to control what I saw, heard and ultimately wrote in ways that could not be acceptable to an academic researcher. Others refused certain requests (e.g. to record on-site, see also note 4) to protect the privacy of their customers. More unexpectedly, documents such as training manuals and assessment criteria were commonly de®ned as con®dential and not to be reproduced, on the grounds that such texts constitute commercial assets from which competitors might bene®t if they were in the public domain.
In addition it proved dicult to interview employees in their workplaces, both because their work routines left little time for it and because of reticence engendered by the culture of surveillance.
When I became aware of these problems I resorted to approaching employees of centres where I had not secured any ocial co-operation, and speaking to them outside their workplaces, without the knowledge of their employers. This approach precluded on-site observations in the centres concerned, but it gave me access to more textual material (employees were generally not troubled by the commercial implications of letting me see their manuals) as well as more extensive and candid interview data. Even so, many of my subjects feared disciplinary sanctions if it were discovered that they had spoken to me and passed on internal documents. I have therefore left them anonymous and used generic labels (e.g. `directory assistance centre') for the centres they work in.
# Blackwell Publishers Ltd. 2000 330 CAMERON
STANDARDIZING SPEECH IN CALL CENTRES: SCRIPTING AND STYLINGThe institutional regime of the call centre exempli®es the hyper-rationalizing
tendency that the sociologist George Ritzer (1996) has dubbed `McDonaldization'. For Ritzer this tendency is de®ned by its drive to maximize four things:
eciency (the most output for the least eort), calculability (the measurement of quality in terms of quantity), predictability (as little variation as possible) and control (of workers' activities by means of technology). Since explaining how these notions apply to the speci®c case of the call centre is also a useful way of describing the workings of call centres to readers who may not be familiar with them, I will examine them brie¯y in turn.
Eciency is maximized in call centres by designing interactional routines so that they consist of the fewest moves needed to complete a given transaction successfully. For example, in the directory assistance centre, the standard routine for processing a request for a phone number has the `core' moves `which name please', `which town', `which address'. This re¯ects the fact that the software used to retrieve phone numbers needs all and only the answers to these questions (preferably in the order just given) to trigger a search. It is also speci®ed in the manual that operators must repeat back to the customer the answer s/he gives to each `core' question. This might appear inecient, since it doubles the number of moves made by the operator, but it is intended to reduce the risk of incorrect details being input and preventing the successful completion of the call.
Calculability is maximized by setting targets for the time taken to process calls, and judging the quality of employees' work in terms of the number of calls handled in a given period (though as we will see, this is not the only measure of their performance). Operators in the directory assistance centre, for instance, are expected to process standard enquiries in 32 seconds or less. Operators in the rail reservation centre are given a target of four minutes per transaction.
The use of standardized scripts for common routines enhances calculability as well as eciency, since the duration of a pre-scripted routine can be estimated more accurately than if there is no script. Though the customer's moves are not scripted, it has been suggested that customers dealing with employees who follow scripts are apt to `routinize' their own behaviour in response (Leidner 1993).
Not all call centre regimes use scripting proper (`scripting' being de®ned here as the provision of a full speci®cation for every word uttered by the operator.) An alternative is to provide a `prompt sheet', which speci®es what interactional moves the operator should make in what order, but does not prescribe a standard form of words. Some centres do not even go that far, providing only general guidelines for the `staging' of a transaction, leaving the exact number of moves in each stage to the operator's discretion. Others use some mixture of the strategies just described. These options exemplify diering degrees of emphasis placed on the predictability of call centre interaction. Scripting maximizes # Blackwell Publishers Ltd. 2000 STYLING THE WORKER 331 predictability, and during my ®eldwork I saw indications that call centres are moving increasingly in this direction, mainly because it is thought to produce eciency gains. For instance, one centre in my sample was in the process of introducing what it called a `standard telephone speech' (a script). While operators were informed that the intention was to improve the quality of service, a memo addressed to supervisors expressed the hope and belief that standardization would reduce call-handling times. It should not be assumed, however, that the only motive for maximizing predictability is to improve eciency. Predictability is often presented as a virtue in itself. Thus a section of the directory assistance centre's manual asks, `why have salutations?'.
(`Salutations' is this company's term for all the polite, interpersonally-oriented formulae that operators are required to insert at various points in the standard routine, such as `thank you', `sorry to keep you waiting' and `just searching for you'.) The answer given is that the use of standard formulae meets customers' expectations of `professional' service by giving them an experience which is `consistent every time they call'.7 Finally, technological control over human operators is seen in various aspects of the call centre regime. Automated call distribution systems dictate the pace of work, while the software used for functions like retrieving telephone numbers, bank account details and rail timetables shapes the sequence and content of many routines. Perhaps the most striking instance of technological control in call centres, however, is hi-tech surveillance. Supervisors can see at the click of a mouse how all members of their team are occupied (in some centres operators who propose to visit the bathroom must key a special code in on their computers so their supervisor can assess whether the time they spend there is reasonable), and they can constantly monitor performance statistics (e.g. how many calls a given operator has taken during a shift and what their average duration has been). In addition, as I noted earlier, the phone system is typically set up to permit `silent listening' by supervisors to calls in progress, and taping of calls for retrospective assessment. These surveillance practices focus more speci®cally on the operator's handling of the interactional task, rather than simply on her/his performance as measured by statistics. If a script is in use, for example, silent listening and taping will be used to monitor operators' compliance with the prescribed wording. But even when there is no script, surveillance is used to monitor various aspects of operators' verbal behaviour. Whether call-handling routines are fully scripted, partially scripted or unscripted, their performance is usually subject to detailed speci®cations of the manner or style in which the operator should interact with callers. This is the approach that I refer to as `styling'.