![[figure showing different stages of aggregation during resource transformation]](Figures/Fig5-1.GIF)
A theory of presentation and its implications for the design of online technical documentation
©1997 Detlev Fischer, Coventry University, VIDe (Visual and Information Design) centre
Presentation is not just operating with existing resources: it simultaneously generates and changes resources and thereby its own context [1]. This generative capacity is here called articulation. It happens both on the substantive and operational level: new material resources overlay existing resources, and their articulation differentiates and re-defines the patterns and protocols on which all presentation relies. In learning contexts, this change is dramatic; in the professional reference context, the pattern is much more differentiated, and changes through new instances of presentation are more subtle.
Articulation can be transient or material and in that it may produce transient or material resources. Transient articulation produces accounts, conversations, and indexing gestures linking material and transient resources. Transformations produce material resources. Transformation happens in writing, drawing, cueing, ordering or re-filing of documents. The resulting documents are material resources of varying permanence depending on the resource substrate.
Permanence influences resource addressability. Resources articulated in the transient medium of speech may be addressable for a just a few minutes; jotted notes or post-its for days, memos and reports for years to come. Some articulations change the resource display, for example, through the annotation on diagrams, while others change the resource structure (and thereby the access protocols), for example, through re-filing.
The confluence of articulation and navigation is the substrate the presentation focus (cf. chapter 7–Confluence).
Even transient articulation shows the high dependence of every contribution on the materiality of shared objective referents. The referent domain exists independently of any particular presentation. Paradigmatic for such substrate is material equipment, and, in the realm of documents, formally binding types such as schematics and service contracts. That these may be interpreted differently only confirms the common referent as shared and objective. On the other hand, users' foci (and their meaning) are not shared, although each individual focus is affected by the confluent foci of other participants (cf. chapter 4–Problem). This is apparent when a presented aspect is misunderstood and requires conversational repair.
However, the materiality of the referent of presentation is not directly accessible by means of scientific measurement or procedures such as the protocol sentences of the positivists [2]: it is a dialectical moment, not ontologically given. That every observation of materiality (e.g., through descriptive protocols or recordings) is subject to well-rehearsed epistemological limitations does not force the conclusion that, since observation cannot specify at the same time its object and its operation, materiality as such has to be refuted [3] —instead, it is the necessary, if inaccessible, moment that ties together the multiplicity of foci and meanings, and thereby affords their comparability. Different meanings would have nothing comparable if not for the fact that they all ground in, and modulate, the objective substrate [4].
Meaning is therefore neither objective-universal nor subjective-particular: it is negotiated in articulation, in a way that different participants sign up to an account of the situation—or differ, but in a way that they can render each other's position so as to prepare or defend a different argument. Semantic differences go beyond details of rendering and interpretation: they are grounded in the ‘understood’ referent domain with its impending actions and assignments of responsibilities and costs.
As articulation aggregates emergent meaning around a problem, it causes the emergence of new problem dimensions. Emergence is afforded by three properties of articulation: production, momentum and peripheral surplus.
Production means that the act of production in speech, writing, or drawing increases the pool of potentially addressable resources. Addressability decays relative to the permanence of the chosen presentation medium [5]. The product may be a resource or a resource pointer.
Momentum rests in the temporal and logical structure of articulation media such as conversation or writing. The grammaticality of language can often lead to conclusions beyond personal expression—others can finish interrupted or otherwise unfinished sentences. Momentum can afford premature saturation, as in turn-takings where speakers interrupt utterances of others as soon as they believe to have grasped the intended meaning. Consequences of articulation however are less serious than in navigation or action, since repair costs are lower (cf. section 3.6 Presentation weight in chapter 3–Context). Emergent articulation is tentative since it affords flexible means of reversing the momentum and correcting errors on the fly. Users may interrupt themselves, uttering alternatives with increased strength and thereby re-casting prior assumptions as merely tentative.
Peripheral surplus means that articulation constantly resonates on a periphery of patterns. Peripheral surplus produces associations, assonances, or marginal attributes of past experiences, all of which may reveal new problem dimensions. The resonating periphery thus becomes available for articulation by the speaker or others. One can interrupt oneself as much as others, to make such potential articulation the actual focus. Elements of peripheral surplus can appear important by kind and yield pointers rather than content; a kind of feature, for example, may appear potentially relevant and become earmarked for future yet unknown questions.
Transient resources are articulated through conversation and gesturing. They are a product of aggregation based on the relevant domain and problem pattern and users' life history which accumulates experience [6]. Conversation is the ubiquitous transient articulation medium. The conversational focus decays quickly as utterances are superseded by new utterances. The momentum of conversation aggregates meaning ‘on the fly’—the speaker (or writer) may not know where and how the sentence might end. Conversation surrounds both navigation and transformation. Sketched conversations in the field and transcribed protocols of dialogical evaluations show conversation as a mutual accomplishment of meaning [7].
Another transient medium is gesturing. Gestures tie users' conversations to material or transient displays: ‘Angus' finger draws an imaginary compressor on my desk’ [FN 10/7/95 P20]. An indexing quality similar to gesturing can occur in remote conversations which trigger situated imagination or guide resource use over the phone [8].
Transient articulation continuously reaches out beyond the focus: it refers back to its aggregated past, and projects and hypothesises its impending future. The past is available as navigable resource, whether this resource is a document or a transient utterance to which users may refer in conversation. In a similar way as navigation assumes the existence of targets to be realised, the momentum of articulation reaches into the future through hypotheses which have heuristic value, i.e. they narrow down the problem space both when they prove to be correct, and when they turn out to be wrong and are eliminated.
Hypotheses are predictions made flexible through the implicit admission of potential error. They limit the complexity of the available resource by focusing on one aspect that seems to further problem aggregation. Hypotheses extend the presentation focus in that they treat the unknown future as potential fact. They thereby act as a momentary bridge for tentative articulation, to be retrospectively validated (or invalidated) in its course. As they fall invalidated, they instantly become the ground for more appropriate hypotheses. Articulation thus generates a highly recursive scaffolding for presentation. Its validated, eliminated, or pending hypotheses reference each other in a sometimes haphazard manner, correcting, extending, and retracting to, earlier hypotheses [9].
Users often articulate hypotheses on little evidence, whether the displayed resource potentially allows more considerate hypotheses or not. It is rare that users carefully scan the entire display before tentatively assigning meaning which then constrains further articulation or navigation. A hypothesis is therefore not a conscious choice from a set of options but rather an immediate resonance effect.
Articulated hypotheses have heuristic value for presentation. They give shape to estimated conditions or consequences in a transient medium, and trigger responses from other participants sharing the presentation. This form of tentative action [10] enables users to test the water before committing to consequential action. In managing a problem in the field, this is frequent in both conversations and informal correspondence. ‘This is a preliminary offer in order to obtain your views on the subject. Should it be acceptable we will make the offer formally to…’ [FN 12/6/95 P47].
Tentative action calls for responses which can act as qualifying or scaffolding resources. They may be sought explicitly, by tapping into other users knowledge, or implictly, by tentatively articulating problem dimensions that invite correction by more knowledgeable users.
Responses are explicitly sought through questions. The metaphor of tapping other people's knowledge frequently appeared in accounts of service engineers. It shows the extent to which remembering is a social, not individual-psychological activity. Instead of having to remember the resource, it is enough to remember another user as resource pointer for secondary navigation which will produce the resource content.
Responses are implicitly sought through utterances which reveal a corrigible state of knowledge, e.g., through hesitation or guesses which implicitly appeal for communicative validation. Such articulations often work in the mode of ‘correct me if I'm wrong’. The greater the power differential between agents, the stronger the tendency that the articulated tentative action is modelled on directives inferred from the responses of the dominant part of the differential. Novice lay users writing their ‘user report’, for example, stuck closely to the merely suggestive outline of questions printed at the beginning of the user report sheet, and voiced only muted or polite criticism. Engineers, on the other hand, often directly criticised the cinegram design at the very moment of ambiguity or breakdown, and suggested alternative designs.
Responses following the tentative action of articulation provide communicative validation which brings users' individual position in line with the position of the group. This is most obvious in cases where situational or operational protocols are ill-defined, as in the cinegram evaluations. Here, users' articulations of planned or possible actions and estimated consequences begged reinforcement, correction, or other subtly qualifying behaviour in other participants (whether these were other users or evaluators). But even in situations where situational and operational protocols are well-defined, users' shared or intermittent articulation affords qualification and communicative validation.
Plans are either already articulate(d) as a schedule in the material resource (like the chain of oil system components engineers check one-by-one for possible leaks), or they are articulated by users. Plans may be articulated in conversations between users (‘Can you go and check please’) or in spoken or silent utterances [11] of individual users' (‘Why don't I have a look at that’). There is no reason to assume a presentation-independent mental substrate of plans, since there is no way of accessing such substrate independent of some form of presentation (as verbalisation, list, algorithm, model, diagram, and so on) [12].
Transformation [13] is the articulation of resources through media that produce permanent resources, such as writing, drawing, or re-filing. It draws on measured dimensions of the referent such as out-of-spec operational values triggering fault indications, and on available resources such as accounts, reports, drawings, or line station logs.
Transformation optimises resources for the target setting in that it combines and filters dimensions according to the form of the produced resource or relevant protocols (cf. section 8.2.2 Resource optimisation in chapter 8–Design). A trouble evaluated as category-A problem by service engineering will trigger a protocol for a mandatory memo, a so-called A-sheet, with a certain distribution list (material transformation), or trigger the building of a cross-departmental task force that will investigate the trouble (structural transformation).
The document or system designer as the author of the transformation hopes to capture the needs of the target setting. The new resource combines, filters, interprets or cues available dimensions according to their perceived relevance for the target setting. It records selected dimensions of things at a given time to facilitate other activities at a later time (or designs the transformation protocols for such measurements). The resulting document makes already recorded measurements available and thereby saves re-measuring. It also provides images and thereby saves imagination (regardless of whether this is mental imagination or physical inspection). The documented price on a produce saves the generation of the price through measurement at the point of sale. Removal instructions save potentially hazardous guesses as to ‘which part comes off first’. Implied is a validation context that secures shared standards of measurement.
The fact that a document can stand in for the referent object or process simplifies work. But transformation severs the link to the referent [14]. Documents can hide complexities which would only be revealed in a recourse to the referent. Opacity is therefore linked to reliability; are there protocols in place which can guarantee that the document tells me the truth? The issue of reliability hinges on the possibility of comparison, which again emphasises the importance of the validation context with its primary and secondary processes that may qualify or falsify the resource [15].
A basic distinction can be made between root resources and derivative resources resulting through transformation. Transformations can be followed back to the root resource, i.e., the material equipment and processes that motivated it. In service engineering, root resources are equipment and users. Users are root resources in that their observations and decisions are used as disputable but irreducible resource for further presentation [16]. Many presentation resources are derivative resources, i.e., results of transformations from observation or measurement of the root resource. Documentation (whether print or on-line) is the particular subset of derivative material resources. It may be generic, as in ATA manuals, or case-based as in ACMS data logs, polaroids, reports, etc. Further types are informal resources and user-designed resources.
Derivative resources are mostly used by exception when a root process breaks down [17]. Users prefer not to read documents if they can avoid it [18]. If certain referent dimensions are needed, documents often seem less attractive resources than informal conversation within a community of users [19] or access to transient resources which provide fine-grained feedback, such as telephone help lines. One reason may be that transient oral resources are less disruptive of primary activities which are themselves not oral, while every document use disrupts the primary activity it supports [20].
Transformation is material when it happens between units of resource, and structural when it involves changes to the resource architecture. Units are utterances, notes, to-do lists, memos, or reports. Architectures are, for example, filing systems which may allow transformation within the structure, for example, through re-filing, or the transformation of the structure itself. System design, seen as transformation, is an example of the latter. Cueing optimises the given architecture by improving the visibility of points of entry and salient referent dimensions.
Unit transformation re-contextualises problem dimensions as it crosses boundaries of resource or setting, human or technical. Transformation may be classified as moving from technical to technical, technical to human, human to human, and finally, human to technical settings (cf. figure 5.1).
Figure 5.1. Different resource aggregates are generated on different levels of presentation. Each new aggregate contextualises the previous aggregate: the ENGINE causes an out-of-spec value which the FAULT INDICATION SYSTEM contextualises as a message displayed on the flight deck or recorded as maintenance message by the aircraft's ACMS [21] system; LINE MAINTENANCE records indications, relevant context, and remedial actions in the line station logbook; if the problem remains, Rolls Royce SERVICE REPS will be involved, who may eventually produce a fault report which sums up the known facts and tried remedial actions. The last aggregate is a trigger reaching the level of SERVICE ENGINEERING.
Technical-technical transformation occurs as interactions of systems on different levels, for instance, when values of a first-order mechanical system trigger signals in a second order indication system, which are then transformed into messages in a third-order display system. Technical-human transformation occurs when system displays are interpreted and contextualised in human renderings, messages and reports. Read-outs of oil temperature and pressure during the flight cycle, for example, are interpreted by service reps in the analysis of engine events. Human-human transformation occurs when users respond to messages by other users. Human-technical transformation occurs when users act upon mechanical equipment, i.e., by devising repair schemes or test procedures that will yield required dimensions such as volumes or flow speeds.
In following the chain of transformations from the values reflecting emergent breakdown to the conceptual level of a problem, one can see a development from the specific and simple to the general and complex and back to specific and simple. Some out-of -spec value turns into messages which turn into indications of trouble; the trouble is then formulated as a problem, analysed and decomposed into hypotheses, questions, and queries. Transformation towards the general increases the complexity of resources and reduces their number. Values on the numerical level are simple and unambiguous, but numerous; messages and indications on the lexical/pictorial level are more complex but fewer. The problem as top aggregate on the semantic level is most complex, but it is just one (or a few). It can be seen as a combination of all emergent reasonings and actions as they begin to shape the ‘case’. Although the problem may have one descriptor (say, the ‘O-ring problem‘—cf. section Sources and coding in App. VI–Grounded theory applied) its complexity means that it has no single resource substrate: it has political, social, economic and technical aspects embodied in many different documents.
Transformation implies that the form of the new resource type selects appropriate dimensions to be articulated in the new resource instance. Examples of resource types are produced by technical systems (such as computer charts logging performance monitoring data) or users (such as line station logs, memos, reports, etc.). Some formal document types literally start with a form: the first page of engineering reports is a boilerplate which can be searched on the STAIRS mainframe system. Examples of resource types in technical systems, e.g., for flight deck indication, are formulas to be instantiated with measured values.
Combination includes dimensions predicted to be relevant or required by the resource type. It implies filtering out dimensions predicted to be redundant. Examples of resource combination are ACMS reports combining a range of faults or out-of-spec measurements, or service engineers adding supportive material such as fault reports, schematics, or copies of manual sections to requests or memos. In technical-technical transformations, protocols for combination and filtering are encoded in the software which defines what measurements are fed into each particular calculation. Measurements of value thresholds, value combinations, and additional conditions are filtered and combined to generate particular messages. A high differential pressure across an oil filter, for example, will trigger a ‘filter blocked’ message, except for start-up conditions where such high pressure differential is expected. In human-human aggregation, reports such as executive summaries for management filter out detail considered irrelevant, or combine a range of problems to support an argument.
The problem of transformation (as of design in general—cf. chapter 8-Design) is that any optimisation through combination and filtering of dimensions predicts what users are likely to know already, i.e., what can be assumed to be provided by the domain pattern activated through display resonance (cf. section 4.5 Problem pattern in chapter 4–Problem). Such prediction works for stable settings such as a stable group of co-operating engineering professionals, but it is already problematic with engineers unfamiliar with a specific design, and potentially dangerous with users that have no engineering background [22]. For example, at the draft stage, the illustration of a scavenge filter in the Trent 700 ETG omitted the bolts holding the filter housing. The designer's rationale for this omission was both based on logic (‘there must be bolts to hold the housing, otherwise it would fall off’) and based on familiarity (‘everyone here knows that such a filter housing is hold by three bolts’). However, a seasoned professional unfamiliar with this specific filter design misunderstood this optimisation. He implied from the drawing that the housing is held by one central screw at the bottom, which is in fact just a drain plug.
Transformation protocols range from simple and predictable to complex and contingent on situation and context. The technical-technical transformation protocol is simple but unaware of the meaning of transformation. For example, it is insensitive to changes such as equipment wear which can lead to an abundance of irrelevant transformations, e.g., the much hated nuisance messages of engine monitoring systems. Technical/human to human transformation is often complex and hypothetical in nature, which means that it involves many informed decisions. This is evident in the note at the bottom of one fault report compiled by a service engineer:
ALL THE FOREGOING IS COMPILED FROM MULTIPLE VERBAL ACCOUNTS AND IS SUBJECT TO AMENDMENT FOLLOWING WRITTEN INPUT FROM OUR REP IN SINGAPORE WITHIN THE NEXT 24 HRS. [FN 12/6/95 P27]
Technical-to-technical transformation is automatic since the transformation protocol defines the type of container, filtering and combination, and many display characteristics. In the case of common aircraft or engine history databases, every fault message originating in a particular aircraft or engine passes through the filter; not, for example, only messages related to a particular type of fault, because that would involve intelligent transformation classifying diverse symptoms and messages under one conceptual heading. Such intelligent filtering happens in the container type of Engine event file, but here, there is a great diversity of document type, format, media, etc., and therefore, no common retrieval mechanism other than browsing and retrieval by temporal order.
Transformation creates reference distance between the resulting derivative resource and the root resource. Reference distance must be distinguished from navigation distance (cf. section 6.6 Navigation distance in chapter 6–Navigation). There are several measures of reference distance, for example, the number of resource steps, combination of resource dimensions in any individual transformation, and timeliness of measurement recorded or implied in the resource. Reference distance affects estimated resource reliability and options for validation through feedback.
Fewer transformation steps mean less reference distance and fewer possible sources of error. Reference distance is zero when looking at an particular part in vivo. It increases as the part is shown on a photograph; it increases further when the part is shown in a drawing, and still further when this drawing is faxed or retraced.
An example of reference distance in human-human transformation is a ‘cascade system’ of instruction. When the DRUID system was introduced in the Service Engineering department, one member of each group was delegated to a training seminar. The participants were asked to pass on their new skills to colleagues. But gaps in the cascade through illness and errors or losses in transformation contributed to the situation that the majority of service engineers either avoid use of DRUID or use it only rarely and shallowly.
The hypothesis is that in presentation, users optimise the resource pool in order to minimise reference distance. Faced with missing or inferior resources, users navigate to better resources originating in settings closer to the root resource, for example, by asking a subcontractor for more detailed schematics, possibly via intermediaries in design departments. Faced with resources such as undifferentiated text, complex diagrams, or amorphous data logs, users articulate selective aspects through cueing (highlighting text, labelling rows and columns of data logs, colour-coding schematics and diagrams) or through transformation into new resources (e.g. visualising selected data slices in graphs). Given the proximity of the root resource, equipment, Rolls Royce service engineers will prefer the real engine to photographs or other derivative resources, even if that means additional navigation effort, e.g., walking down to PTF, the production and test facilities.
Transformation oscillates between two activities—measuring one resource (root or derivative) and informing another resource. Also, there are further interacting secondary activities, such as cueing resources, handling tools, or accessing help or support systems. This oscillation opens the door to error, particularly if the transformation protocol is complex. As a consequence, documents may not correspond to the things or facts they record, for example, through data loss, quality decay, fraud, or by being out of date, which introduces the need for validation protocols, regulatory regimes and standards.
Examples of complex transformation are users' inferences of flow directions and possible state changes from a static flow system diagram; or, in resource design, the task of deriving a 3-D illustration of a complex component from a number of schematics. Sometimes, more transformation steps can reduce aspects of reference distance, e.g., by filtering data and thereby optimising display scope. Examples are data visualisation of selected slices from engine monitoring logs, or functional optimisation of cross sections in ETGs.
Presentation of problems involving inaccessible parts of equipment often relies on estimated data from models hypothesising hidden system behaviour on the basis of measurable data and domain generalisations. Such transformation into the hypothetical realm means data are only best guesses and often need confluent human qualification (cf. chapter 7–Confluence).
Transformation can impede presentation through noise and mandatory pointers in the display. Noise means that the abundance of irrelevant units or referent dimensions impedes detection of relevant units or dimensions. It increases the display real estate space and may thereby force segmentation and consequently, navigation to hidden views (whether printed or electronic pages). Mandatory pointers in the display are pointers such as problem codes which must be followed up before the relevance and weight of their target can be assessed.
An example from British Airways' Base Support Shops shows the impediment to use caused through noise and mandatory pointers. The transformation protocol of the Boeing 575 engine monitoring system produced print-outs containing hundreds of nuisance messages. The result were severe presentation problems for maintenance mechanics who had to sift through the document in search of potentially relevant messages and had to look up unknown problem codes since they had no means of assessing their importance.
Cueing qualifies the material resource by adding users' perspectives. This happens, for example, through annotation, marking, and underlining of documents. For documents with poor internal differentiation such as performance monitoring print-outs, cueing highlights tokens as addresses for noticing to reduce search time and thereby navigation distance. Some of these cues facilitate switching between views; other cues strengthen the confluence of dimensions in the same view, for example, by circling critical values. Such cueing flips over into user design when engineers decide to render selected data slices as a graph in order to improve the visibility of trends (cf. subsection 8.1.1 User requirements in chapter 8–Design).
A particular case of cueing is the allocation of descriptors which improve grab and visibility of resources as they appear in conversations or displays of pointer collections such as search hit lists, lists of contents or indices. Authorised design allocates descriptors to content according to formal protocols. For example, Rolls Royce's Configuration Department manages the complex number space to secure consistency and uniqueness of numerical descriptors. The descriptor may be the document title, ATA reference, task number, or technical report or engine ID number.
Descriptors have boundaries of applicability. Different settings and organisations articulate different descriptors for the same referent (the engine maker's ‘engine’ is the airframe maker's ‘powerplant’).
The most important descriptor for presentation is the problem descriptor that is articulated as a case begins to take shape. The problem descriptor is important since it calls up appropriate displays and rememberings by activating the problem pattern. Users' informal problem descriptors have much grab: they are usually short, colloquial, hyperbolic or funny (e.g., ‘bean can problem’).
Articulation qualifies and scaffolds the navigation of resources, which is the subject of the next chapter.
[1] The emphasis on the generative nature of reality goes back to phenomenology and ethnomethodology ( cf. Schutz 1972/1932; Garfinkel 1967; Coulter 1979) which tends to play down the dialectic between emergent and pre-existing structure.
[2] Cf. Neurath (1979/1934).
[3] This is the position of radical constructivism ( cf. for example, Maturana & Varela 1987; von Foerster 1984,1993; Luhmann 1992) which draws on concepts from technology and biology such as virtuality and autopoiesis.
[4] A similar argument can be found under the heading of transcendental realism. ‘For to say of two theories that they conflict, clash or are in competition presupposes that there is something—a domain of real objects or relations existing and acting independently of their (conflicting) descriptions—over which they clash.’ (Bhaskar 1989 p18)
[5] The term medium is here employed to mean the mediating process, not the carrier or apparatus. Cf. Salomon (1979 pxviii-xix, quoted after Pinnington 1990 p34) who states that ‘the essential differences between media are distinguished by the way in which they structure and convey contents and not technological media differences’. The medium of television, for example, is the emission, and conversion into moving pictures, of a signal from one sender to many receivers, not the television set. Equally, conversation is the medium which renders meaningful singular utterances or speech acts. Language is not the medium, but the structured repertoire of potential meaning, which is galvanised and put to use through dependent acts of articulation as they become possible and meaningful in the medium of conversation.
[6] Cognitive theories of instruction devise procedures that attempt to turn brains into documents—or ‘learning outcomes’ ( cf. Gagné 1985; Driscoll 1994 p333). But how does the learning outcome cope with incoming problem situations? ‘The strength of…recommendations [of procedural models of instruction] is wholly dependent upon the extent to which they account for all relevant variables and the extent to which the correct relationships among these variables have been identified’ (Richey 1986 p23) Most situations however contain novel or ill-defined variables with emergent relationships. Even if some declarative knowledge is hammered in successfully ( cf. Gagné's ‘nine events of instruction’) there is a lack of transfer to ‘real situations’ (Driscoll 1994 p367/369). Theorists should acknowledge that brains do not ‘store’ data in the same way as folders or filing cabinets—and some do ( cf. Boyd & Pask 1987; Johnson 1991 p139; or Brown & Duguid 1992 p175). ‘If memory were a data storage system it is easy to show that in order to account for what we know each of our brains should be the size of a sphere about one mile in diameter packed with nerve cells. However, when of this size, the operation to recogize, for instance, the presence of a lion in its field of vision takes the brain about ten years’ (von Foerster 1984 p216).
[7] Ethnomethodology has made an important contribution to the research of the emergent and mutually constitutive nature of conversation (Cf. Garfinkel 1967; Coulter 1979; Suchman 1987). However, by focusing on individuals' situated production of meaning, ethnomethodology ignores the grounded nature of conversational action—its consequentiality which validates meaning, often with considerable delay, as it turns out to conform or clash with features of structured society. Such delay is excluded qua method, in that the analysis concentrates on emergence in small samples.
[8] In one observed case, an engineer received a
call and immediately asked the customer to retrieve and look at a certain
ETG. The service engineer himself looked at the ETG displayed at the wall
behind him, and remotely directed the customer's navigation:
[S. swivels round and looks at his ETG at the wall] ‘You need to go
to the internal gearbox picture–go to the middle of it– it's the outer
race of the inner bearing, it's that one there–change that, the only bearing
on the radial shaft…’ [FN 2/8/95 P6].
[9] In the following example, articulation produces
a series of hypotheses that grope for an answer to the question ‘Why are
there two filters in the oil system?’:
‘It could be because of failure, if one clogs up…then at least you've
still got a second one that will—ahm—filter out the system…although that's
only if they are working in parallel, if they are working in series it
could be because one will take away the larger contaminants and one will
take away the smaller contaminants afterwards…why do you need two?…could
be because one was working at a higher pressure than the other one, which
it obviously is…’ (samples of a line-by-line analysis relating to this
excerpt can be found in section Sources and coding in App. VI–Grounded
theory applied).
[10] Freud generally treats trains of thought as Probehandeln (tentative action).
[11] Introspection shows the evidence of silent utterances, or ‘experientially distinct entities that exist in the mind’ (Shanon 1993 p308). There is however a qualitative change when silent utterances are ‘turned loud’ in evaluations requesting concurrent verbalisation. The change of medium from imagination to speech increases resonance: ‘Subjects in the verbalization condition committed fewer errors and consumed less task time than subjects in the silent condition’ (Wright & Converse 1992 p1220).
[12] This is the reason why Payne (1991, quoted after Landsdale & Ormerod 1994 p176) found it impossible to decide if users who were asked to explain their ‘mental model’ of a bank's cash dispenser constructed the model in the light of the question, or accessed a genuine pre-existing mental model.
[13] Transformation is related to mapping ( cf. Norman 1986 p32; Winograd/Flores 1986 p85; Weir 1991 p6). Mapping describes the intended fit between the things and processes in the domain, and the entities in the designed object.
[14] This appears not to be the case in real-time monitoring such as in automated flight decks. However, the amount of transposition between raw data and flight deck display shows the same problem of opacity: can the automated flight control system be trusted? Recourse on raw data is often discouraged or impossible. Cf. Funk et al (1996).
[15] For more about the notion of repair-oriented design, cf. Winograd & Flores (1986, p150/165) and Norman (1988 p140).
[16] Users are not only more or less reliable functions within the operational process, but make hypotheses and take decisions that become the root of further events. An example is quoted in a service engineer's hand-written summary of an engine-related problem: ‘Against…maintenance advice, pilot chose to turn back rather than complete remaining 4-5 hrs of flight (quoted concern over possible windmilling damage with no oil)’ [FN 12/6/95]
[17] Cf. Winograd & Flores (1986 p5).
[18] Document designers have realised that documentation may be superfluous if equipment functionality is made more evident ( cf. Horton 1993).
[19] Cf. Mirel's (1988, p284) investigation into the use of in-house manuals in an organisation.
[20] ‘The very presence of documents, be they print or on-line, presents the danger of the user becoming disengaged with the learning/doing processes, because the documents can draw the attention away from the activity at hand’ (Johnson 1991 p139).
[21] ACMS = Airborne Condition Monitoring System
[22] The perspective of emergent presentation suggests that no document module can be strictly self-contained in the sense defined by Walker (1988 p118): ‘A module is a component that says something self-contained and comprehensible. Well de-signed modules would have few connections to other modules; the ones that they do need (to prerequisite and subsequent modules) would be explicit’.
[23] The competitive nature of business is a hindrance for presentation: subcontractors often deliver incomplete or inferior documentation since they are reluctant to give away the blueprint for outsourced components.
[24] As the reason for this the manager quoted ‘purism of the avionics engineers’ who had implemented messages of the type ‘Give us a message if this component works or not’ simply because this was easy to do. BA eventually issued a 50-page meta-document explaining the nuisance messages. I doubt that this additional document made the task of trundling through the problem codes any easier.