Chapter 3    
Fieldwork and ethnographic research


3.1   Introduction

Fieldwork is the basis of ethnographic research. If computers are to become a important tool for anthropological research, the use of the computers must begin during fieldwork. Otherwise, as Davis suggests, 'striking savings in time and accuracy are achieved only after a high initial cost in preparation.' (Davis 1984b: 308). So long as data is recorded in one form, only to be transcribed to another, then preparing field data for computers competes with other tasks, such as indexing notes, transcribing audio tapes, and deciphering genealogies, and can potentially be the most time consuming of these tasks. But, '[the costs of preparation] diminish as you plan and as you use the machine for more jobs.' (ibid:309). The larger tasks might be the initial incentive for computer use, but the benefits emerge from the multitude of smaller jobs,  which  would not themselves have justified the cost in time and money. One of the ways to diminish the barrier of preparation is to use computers throughout the research process.

Although 'striking savings in time and accuracy' are important, these are not the most compelling arguments for using computers in anthropological research, in the field or out, as Kippen remarks:

I believe that the enormous potential of computers to improve the capacity of anthropologists to gather, store, and analyse information has yet to be demonstrated. Clearly, it is necessary to look beyond word processors and databases to other systems, such as knowledge-based systems, that have the power to change the ways in which we as researchers operate.     (1988b:318)

Our most important goal in using computers as a tool for research must be to do better, not simply more, anthropology. If you are satisfied with the state of ethnographic research, there is little purpose for introducing the additional cost and time for learning. The best way to introduce computing into our research is to first replicate what we have done before, but greater benefits will come when computers are used to do things we could not do before, not only for the amount of time these would have involved, but because these could not easily have been conceptualised prior to opportunities the computer as tool can facilitate.

3.1.1  Ethnographic data - how conventional?

In Figure 3.1 I have reproduced a schema Ellen presents for classifying 'permanent' written field records (1984:282). I take his account as representative of the way anthropologists present field records and their  guidelines for producing and maintaining these records, a view in concurrence with Sanjek (1990a:235). But, in practice, just how conventional is this schema? For example, consider the category Ellen refers to as 'conventional notebooks' (1984:282), which I take to be, more or less, fieldnotes (although Ellen seems careful not to do so). 

Jackson (1990) conducted systematic interviews of a non-random sample of seventy users of fieldnotes, sixty-three anthropologists and seven other social scientists.  She encountered considerable variation in what are regarded as fieldnotes (at least for her sample; anthropologists on the East Coast of the United States). She reported that

What respondents consider to be fieldnotes varies greatly. Some will include notes taken on readings or photocopied archival material; one person even showed me a fieldnote in the form of a ceramic dish for roasting sausages. Some give local assistants blank notebooks and ask them to keep fieldnotes. Others' far more narrow definitions exclude even the transcripts of taped interviews or field diaries. … Clearly, what a "fieldnote" is precisely is not part of our profession's culture, although many respondents seem to believe it is.    (1990:6).

In her fuller discussion it becomes clear that there are two main sources of variation. Firstly, precisely where, in the classification of field material, does any particular kind of information belongs. This is a matter of naming rather than of substance, possibly broadening our inclusive definition to the entire range of written records and beyond (and perhaps accounting for Ellen's caution). The second source of variation relates to whether the respondent placed value on, or even recorded field notes, entering entirely into the domain of what Davis refers to as 'Intuitive knowledge' (1984b:304), Ottenberg  as 'Headnotes' (1990:144) or, as Jackson quotes one respondent, 'I am a fieldnote'. (1990:21). Jackson also describes  a range of 'personal' attachments and a mystique which many of her respondents ascribe to fieldnotes (ibid).

Seidel (1991),  author of The Ethnograph (1988), a popular program which supports the analysis of ethnographic texts, discusses conflicts he has reconciling promoting the use of computers for ethnographic research, while being aware that his program is used by some researchers as a tool to do ethnographic analysis, rather than a tool to assist ethnographic analysis. They become 'distanced' from their data by defining ethnographic analysis as those tasks the program performs (1991:114-115) or taking refuge in the 'deification' of data  (ibid:112-114).

The writing and use of ethnographic field material can apparently go astray with or without the involvement of computers. Using computers cannot solve the problem of patchy notes (although they can be used to assess how patchy the notes are (Fischer and Finkelstein 1991)), and are unlikely to assist those who depend entirely on their memory (but see §8.2). For some people computerised notes may not be able to supply the mystique of notebooks and they may not be able to develop personal attachments to computer data files. And it is possible that a program designed to enhance access to fieldnotes might provide an avenue for abuse.

It would be unwise to dismiss these problems. A major consequence of using computers for field records might be to make us aware of these peripheral aspects of producing and using these records. One of the things that applied anthropologists are forever telling organisations is that regardless of what their planners, accountants and managers say the structure of the organisation is supposed to be, it is better to base policy on the organisation as it operates rather than how it is supposed to operate, translating the results back into organisation-ese if required by company reports. The failure of many a computer system situated in an organisation has been due to the inability of that system to cope with actual practice.

But is anthropology and the conduct of anthropological research a fiction? There are many critics who appear to tell us so, although their criticism and our consciousness of that criticism tends to reduce the fictional danger over time. We must not lose sight of an important disciplinary assumption. We may take it for granted that the ethnographer and the ethnographic texts are intertwined to the extent that others may not make much out of the texts without the ethnographer. We may disagree about what is being unveiled.  But we must also assume that the resources the ethnographer uses in the retelling of the material meet some general standard. If Ellen's presentation is indeed a fiction, we should hope it is a useful one.

3.1.2   Organisation of the chapter

This chapter presumes that a computer will be available in the field, where issues differ somewhat from the study. If field access to a computer is not possible (or the field research has already been done), then, of course, both preparation and analysis can take place at your home site. In any case, it is important, in advance when possible, to match intended forms of processing data with plans for the collection of that data. In ethnographic research this is not always possible, nor in all cases desirable, but certain methods of analysis are simply not possible without specific data. 

It is certainly not my purpose to say what you must collect or how you must analyse it. (This would be pointless because I cannot anticipate all my own research needs.)  Any specific methods given are illustrative rather than pre-scriptive. In a research environment methods must come from the researcher.My intent is to inform, to identify what resources are available to work with conventional categories of data, and to suggest how some lesser-used categories of data might become more important. Unlike archaeology, linguistic anthropology  and physical anthropology, the literature for social anthropology is rather sparse in applications of computers to data other than fieldnotes and other texts, survey data and historical documents. However, all means discussed are accessible,  and have been used in the field by anthropologists, although some require more expertise than others..

3.1.3   Past and present feasiblilty of field computing

Early efforts by Weinberg and Weinberg_2 (1972), and by Brown and Werner  (1974) were attempted by shuttling information in and out of the field site for processing, but this was, understandably, not persuasive. The rapid development of microcomputers from the late 1970s - increases in power in conjunction with reductions in size and cost - has made the use of computers in the field not only practical, and increasing numbers of anthropologists have taken computers into the field since the late 1970s. (Tomajczyk 1985; Sutton 1984; Powlesland 1986; Guillet 1985; Dyson-Hudson and Dyson-Hudson_2 1986; Ellen and Fischer 1987; Agar 1983; Case 1984; Werner 1982; Chagnon 1982).

These efforts were not without the drawing of blood. Many of these systems weighed over fifty pounds, and upwards of one hundred with accessories, power supplies and spares. Equipment designed to operate in laboratories or homes at temperatures between 15C to 27C with relative humidity between 40% to 80% (non-precipitating) found itself in temperatures from -10C to 50C and relative humidity of 100% (precipitating), with wily anthropologists luring insects from the keyboard with peanut butter (Dow 1987). No one who has participated has suggested it was not worth the trouble, but that it was trouble no one would deny.

Thus, for most anthropologists taking a computer to the field has become feasible only recently, not because it was technically impossible, but because the logistical difficulty of operating or accessing computers in most field situations was unacceptable to most anthropologists. 

Most of the problems with hardware are, if not solved, under control. There are battery-powered 'notebook' computers the size of a A4 book, weighing less than three kg, and capable of almost anything larger 'desktop' computers can do. A4 size 'notepad' computers weighing somewhat less than 2 kg, can perform complex tasks and store a year's fieldnotes, as can their smaller brethren, the 'palm top' computer, weighing less than three-hundred fifty grams, with 256k-512k RAM, powered for six weeks on two AA sized batteries, and a non-volatile storage capacity equivalent to one thousand A4 pages on a removable (and replaceable) twenty gram wafer solid-state 'disk'. Interfaces for direct entry via hand printing with a pen are available, which work much like conventional paper pads, down to the ability to draw in the margins. Although initially more trouble, small hand-held machines with a one-handed 'chord' keyboard can be used for note entry, with blind text entry at speeds at least as fast as conventional typing, and which can offload the text to a larger computer (say a palm top) for analysis. In short, an anthropologist can carry into the field more computing power  and storage in a three-hundred and fifty gram package the size of a Prince Albert tobacco can than was available in 1980 using a desktop microcomputer.

3.2   Computing aspects of written data produced during fieldwork

Most anthropologists will find written records (Figure 3.1) are the easiest to work with using conventional computing techniques, at least in the sense of the amount of knowledge about computers required for the job. Almost any computer, of any size, can be of some benefit to this process. In this section we shall look briefly at the requirements of ethnography and how these can be reconciled  with computer-based tools appropriate to working with written materials.

Ideally, in doing ethnography we would not be restricted in any way in terms of the information we record and our subsequent access to this information. In practice the very tools which are used (the means of writing and organising this writing) appear to encumber the process, not only because of the time required to create a record of our observations and experiences, but also because the volume of writing becomes so great that it is difficult to keep an overall grasp of what has been recorded, much less the detail. 

Part of this problem is our human inability to remember in a non-selective manner (D'Andredae 1973) , and part is technology, the pen, paper and writing we use to improve our ability to remember. You can use computer applications to address the technological problems as a supplement to or as a partial or total replacement for our conventional technology.

In this section we examine the issues of data planning, collection, maintenance and minimal analysis in the field. Some processing and analytic issues are discussed in Chapter 2, and fieldnotes are discussed in greater detail in Chapter 4. It is unlikely that most anthropologists will want or need to use all of these methods in any one field project. This is a discussion of computing methods available for, and appropriate to, the field, rather than what one should do in the field. In each of the entries basic issues are presented, the kinds of software necessary to address these issues, and, where necessary, an outline of the hardware requirements as these appear in 1993. Guidelines are intended as just that, and it is essential that you seek specific advice from a computing support person or further reading. Software issues are, in part, based on conceptual and practical issues relating to the problems addressed. Hardware is simply a means of implementation, and is subject to rapid changes in capability and price. I have tried to limit speculation to hardware developments that more or less inevitable in a time span of two to three years.

3.2.1   Types of written records produced during fieldwork

In terms of 'conventional' ethnographic research, most data 3, and the most important data, is collected in the form of written records. Ellen identifies seven main types of written records produced during fieldwork:


These forms of data, while not exhaustive, serve as a basis for discussing computer  applications in the field, which must necessarily be oriented towards the overall fieldwork process. 

In general, computers adapt very well to assisting in the collection and analysis of written materials, in part because of the activeness of humanities researchers in devising methods of working with historical and literary texts, and in part because of the commercial 'office revolution' in the 1980s which resulted in the widespread distribution of quality computer-based tools for producing and accessing texts, as well as the provision of tools for representing complex information and facilitities for making reports based  on that information. Some of these tools can be applied directly to our material, with little accommodation on our part, some tools require that  we structure data in ways which are otherwise unnecessary, and some tools require an entirely different conceptual basis to use effectively.

3.2.2   Temporary notes

Temporary notes are, currently, the least likely data to be entered onto computer in a form other than their contribution to the other types of written records. Most temporary notes are written into notebooks, and being temporary, there is little point to re-entry just for the sake of reproducing the notes. Some researchers, myself included, enter the temporary notes into a wordprocessor verbatim and then expand these, but do not systematically retain a verbatim computer copy (cf. Pfaffenberger 1988:33).

You can, of course, enter the temporary notes directly into a computer. Many ethnographers find taking notes on paper distracting to both themselves and their consultants. It is unlikely that a computer will be less so. In many cases there is so little 'connected' text (as opposed to telegraphic prose, outlines, charts and drawings) involved in a days temporary notes that there may be little value added if the presence of a computer is likely to detract at all. In my own fieldwork I have recently been taking some notes on a small handheld computer which is as unobtrusive as a pen and notebook (once I have initially demonstrated it to my consultant). It saves no time what-so-ever if I transfer these notes, unaltered, to my larger laptop computer later in the day, because the time required to set up the file transfer hardware and software takes about as long as typing the notes in (although with far less tedium). The major advantage is that I can partially fill the notes out on the handheld computer at odd times during the day, wherever I might be, which does save time, since otherwise I would be writing the expansion in the notebook, and retyping the expanded version later in the day. Also useful, I have all previous fieldnotes and card indexes (from current and previous research at the same site) on the handheld computer as reference for writing the expanded notes, as well as for preparing visits and interviews.

Two kinds of specialised han-held computers may better justify the direct entry of notes onto computer compatible equipment. The first are small, relatively inexpensive, 'chord' keyboard entry  computers, although I know of no anthropologists using these. These weigh only one to two hundred grams, and are operated using one hand and a small number of keys. Special key patterns have to be learned which correspond to letters. Some people have little trouble with this, achieving relatively high rates of speed in a few weeks. Others find it very difficult.  For successful users, notes can be entered faster (40-80 WPM) than most people can write. More important, notes can be taken in any position, even with the machine in a pocket, matching Margaret Mead's alleged ability to write notes in her handbag. Once in the machine the notes can be edited and read directly from the display, or transferred to a larger machine for expansion into the day's fieldnotes.

The second development is of small, powerful and relatively expensive hand-held computers which accept input in the form of hand printing with a pen. As of 1993, these are too error prone to replace pen and notebook in ethnographic situations if scrawl is most likely to describe the standards of penmanship. However, if you can print fairly neatly, the better of these can record copy which is good enough for rough notes. Indeed the best of these pen driven systems have abilities of contextual linking to contents of other notes and databases which suggests that, properly prepared, such a system marks some some steps towards a process of automatically filling the notes out with personal details, location details and time reference information, as well as automatically posting reminders to follow up issues so marked in the notes. These systems also support the drawing of diagrams, and linking these to textual and database references. With more development these pen-driven hand-held systems have the potential to serve as the complete hardware base for field-based anthropologists.



1.    Temporary notes.
2.    Conventional notebooks, general and specialized according to topic or kind of record.
3.    Diaries, personal and register of events.
4.    Card indexes, such as a name directory or dictionary entries.
5.    Questionnaire returns.
6.    Special records, e.g. register of taped material, special survey material, genealogical charts.
7.    Maps and diagrams.

Table 3.1   Types of written record produced during fieldwork  
        (Adapted from Ellen 1984:282)

 3.2.3   Conventional  notebooks, diaries,  and interview  transcripts

These materials are suitable for computer-based entry in the field, using standard wordprocessors (§1.3.1.1),  hypertext editors for entry (§2.3.4; §4.6.5) and/or simple textual database managers (§4.6.3) which are easily adapted by a casual user for these purposes (§2.3). These are also the data from which much post-fieldwork benefit will come, since the preparation time will be considerably lessened for data which is among the more tedious and which requires the most sensitivity and specialisation of (anthropological) knowledge to enter; information which should usually be entered by yourself on return. See Chapter 4 for a review and discussion of principles and methods of fieldnote entry, management and analysis.

3.2.4   Card indexes

Card indexes are a common cellulose-based information technology (CBIT :- for storing and accessing a variety of kinds of data in the field, including word lists, bits of information on individuals, and names of objects, plants and animals. Among the advantages of card indexes in the field are flexibility in inserting new material, and the option of sorting the cards into different orders for different purposes. Card indexes are particularly easy to implement using conventional computer programs. The type of program you should use depends on the kind of information, and how it is structured on the card equivalent. 

Notebook computers are capable of dealing with any of the the options given in the following discussion. If your field situation requires very low power use, notebook computers capable of operating several weeks on standard batteries can used for most of these options, if on a smaller scale, but probably adequate for a season's fieldwork.

To simply  replicate the function and operation of a manual card index on a computer is very simple; you can enter the information using a basic text or wordprocessor (§1.3.1.1), and use the 'Find' operation of the editor to locate sections of the document by doing a word search, and, if desired, copy selected material  into another document. This has the advantage of using a basic tool for two purposes, entry and retrieval, and permitting you to lay the information out in any way that you like which is compatible with the word processor or text editor. Given that the more recent word processors have the capability to store and display high quality images, and even video and sound (§5.3; §3.3.2) this can be a flexible structure indeed. However, there are severe constraints if your document becomes too large (and thus slow) or if you want to process the information further. 

If you want to automate storing the results of a word search, most text-based computers  have at least one program which will accept textual keyword(s) and search through your text-file  created using a text editor, displaying on the screen or printer, or storing in another text file, the lines in the source file which contain the keyword(s). If your card index is simple in structure, word lists or simple inventories, this may be adequate  If a single line is not adequate to represent the contents of your card index application, there are other programs which will work with larger sections, separated by some distinctive marker you  type in after each entry. If you have stored the output of these searches in a new file, other programs can be used to count the number of lines or to sort these files into new orders.


For more complex data processing requirements see §2.3. In particular, for fieldwork purposes consult §2.3.4 and §4.6.5, which describe uses of hypertext editors. The principal advantage of hypertext for fieldwork is the extreme flexibility for representing different kinds of textual and non-textual data, together with flexibility in changing the design as your work progresses.

3.2.5   Questionnaire and survey returns

Software for the analysis of survey data is perhaps the best known application of computers to social science research, for many the only known application. There are a broad range of applications for statistical analysis, questionnaire tabulation and generating tables for examination or publication. The kinds of program you require depends on the complexity of the survey instrument and the size of your intended sample. For surveys with mainly numerical data and where the number of cases is a few hundred, a spreadsheet calculator may be adequate (§2.4.1). For more complex surveys, with linked or optional questions, a statistics program will probably be required. If the survey is especially complex, you may in addition need some kind of database management system (DBMS) (§2.3) which you can use to create simpler data structures for the spreadsheet or statistics program.

The value of dealing in the field with questionnaire and survey returns depends on the use you intend for these while in the field, the sensitivity of their content and the means you have available to enter these into a computer.  A problem, both in the field and out, is the preparation of survey data for the computer. If you intend to simply look over the questionnaires in the field, and you must enter these yourself from the paper forms, then it may not be a justifiable expenditure of valuable field time. If you think a particular questionnaire might be useful for planning further research while in the field, then it may be worth the time, if you have included an appropriate spreadsheet or statistical package in your software toolkit. In many cases for planning field research an analysis of a relatively small sample (< 50) of the questionnaires or survey instruments will probably be adequate.

In some cases it may be possible to have someone else enter the questionnaires into your computer, or another which can produce compatible files. For example, in many countries banks and public utilities have spare data entry capacity, and will enter the material for you at a reasonable rate.  If the questionnaire is a long one or covers more than fifty cases, it is a good idea to pay for it to be entered twice, and use a program which compares two files to identify errors (which are likely, if not certain), or pay for this service as well (which is always available if they take in outside work).

There are also various pieces of equipment which can be used with forms to 'automatically' enter forms from surveys which require mainly yes/no or multiple-choice responses, using some kind of mark-sense form, most commonly manifest as a computer card similar to the old punched card. The response is recorded on a special form, marking with a pencil in a particular position, and directly read with a machine. This is an expensive but simple solution. You must get such a reader with software and hardware interfaces which suit your machine, and test it for accuracy before leaving for the field, and take along a suitable supply of special forms with you. Another option which can be used with forms on which choices are recorded on plain paper are optical character readers which can read carefully hand printed numbers, or if you know a programmer, a program can be written which will spatially interpret the marks on a form which has been entered using an image scanner/digitizer  (§5.5). 

If you (or your field assistants) are filling in the questionnaires in the course of  interviews,  direct entry of the questionnaires into small notebook or hand-held 'palm top' computers can be a productive option. A form representing the questionairre can be 'programmed' into the small computer and entries made and recorded quickly.

Green (1988) developed a program based on an expert system (§8.2) to assist in converting textual material such as interviews into questionnaire type data, by having the anthropologists develop a set of rules for answering the questions from a transcript and encoding these in the program. The program then asks the questions for each case, providing standardised assistance in classifying each case with respect to the question. This has the advantage that each questionnaire is produced in a consistent manner from relatively unstructured data.


Input File (Urdu/Panjabi)    Search for    Output

admi, man    boy     bucha, boy
aurat, woman        lerka, boy
bucha, boy
buchi, girl    lerk    lerka, boy
...        lerki, girl
lerka, boy
lerki, girl
log, people

Figure 3.2. Results of two keyword searches of text file representing Urdu/Panjabi word list.

3.2.6   Special records

Ellen (1984: 282) includes in this category materials such as registers of taped material, special survey material and genealogical charts, and most other kinds of written data of a 'special' sort.  In the strictly written sense, most of these can be adequately dealt with using either textual databases or more structured database management systems as in §2.3, possibly in the case of genealogical data using the programming language Prolog (Chapter 6), and for special surveys which incorporate visual and aural materials a hypertext application (§4.6.5; §2.3.4).

For many kinds of special records such as these it is sometimes useful to write a dedicated application for entry. This can be done with fairly limited programming skill using a hypertext editor program, most database management systems, or you can specify the function of the application to a programming partner.

As an example, for genealogical material  a simple graphical editor for entering relationships directly in the form of kinship diagrams, and attaching information about the household members with the kinship symbols on the screen. The program illustrated in Figure 3.3 was written using a common hypertext program (Hypercard) in about two hours, most of which was spent making the icons for the different symbols. It underwent a slight modification after some suggestions in a workshop to improve the control in manipulating the symbols, taking another 20 minutes. 

Similarly, the application in Figure 3.4, written using the same hypertext program, is dedicated towards making a register of the contents of audio and video tapes. It simply is a form for entering textual notes and tape index counter readings which can be used as I am listening to the tape, and for adding cross-references afterward. Not a general purpose application, but it took only about twenty minutes to originally construct, with modifications added as needed while in use.

In both of these applications the data can be later output into more conventional forms if required.

3.2.7   Maps and diagrams

Most anthropologists seem to generate a lot of diagrams or other graphical representations while in the field, if with varying degrees of skill. For some kinds of research diagrams are essential; as a illustration of ethnographic detail (the placement of a decoration in a house), to denote certain kinds of relationships (genealogies) , or for creating representations as an interviewing aid (drawings of material culture). See §5.2 for further  information on drawing using a computer, §3.3.1  and §5.3  for computer based photographic and video images, and  §5.4 for further discussion of maps.

3.2.7.1   Applicability of diagrams to fieldwork

Although the use of computers for making diagrams is of great value, especially in the presentation of ethnographic material, its use in the field very much relates to the role of drawings in your own fieldwork. 

 If rough diagrams begin life on paper, there may be little point to converting these to rough diagrams in the computer, except for the possible benefit of integration into your fieldnote database.You might, of course, want to catalogue referenes to these using a database. If you restrict yourself to rough diagrams, however, the effort is not very great. If you include an image scanner/digitizer (§5.5) in your hardware toolkit, the effort is rather less since you are not drawing but copying a drawing onto the computer. The latter will usually be edited using paint type programs (§5.2.2), although there are some applications which will attempt to convert these for draw type programs (§5.2.1). Quality problems in the field are of rather less importance than for published work. 

Even with special software and hardware for this purpose, it takes considerable time and effort to prepare a good diagram for publication or lectures, and of the many rough diagrams you are likely to make in the field, only a few will be selected for this purpose. High quality drawings should be done after leaving the field, unless you have a particular purpose in mind.

I use a limited number of drawings as interview aids when discussing material culture, and find this very useful. From a printed copy I elicit 'corrections' to my drawings, helping me to separate etic detail from emic detail, I attach the various responses to the computer-based drawings (in a hypermedia program §3.2.6), using these as one means of access. It works well for me, especially given my limited skill in freehand drawing with pen and paper. 

3.2.7.2   Special purpose graphic tools

There are an increasing number of programs which employ graphical tools potentially useful in fieldwork. These include programs for drawing and manipulating maps, programs for direct genealogical data entry in the form of diagrams, such as that described in §3.2.6, programs which can assist in the building and testing  of taxonomies, based on recent work by geneticists on cladistics, and graphic editors for the direct input of musical notation, as well as facilities for playing back music in this form.

Maps, even rough ones, have considerable utility when on a computer. Drawing a good original map is difficult, even on a computer. But the sorts of rough maps (in appearance, not necessarily in accuracy) that are commonly used in the field are relatively easily entered (see §5.4).Use of maps in the field include monitoring cultivated areas, plotting social groups and kinship links, or other cultural variables, against the map, and tracking the coverage of surveys and questionnaires. Using a printer, the map can be reproduced in a variety of sizes, 

Figure 3.1  Hypertext kinship input form

Figure 3.2  Hypertext videotape register

with different textual legends as required. Maps are discussed in more detail in §5.4.

The value of graphical input of data in the field depends very much on your research. The strength is that the data input format (and thus at least one database record) is the same as the report or output format. If you tend to record genealogical relationships in diagrams in the field, then a genealogical editor may well make sense as a means of transferring data from notebook to computer, even more so if you intend direct entry working with a consultant. The graphical structure of the data represents structure as well, and this can help ensure that you have included everyone, and recorded all the relevant data. On the other hand, if you mainly record genealogical relationships in textual form (recommended in any case if you are recording genealogies of more than twenty or so people at a time), and mainly want to be able to see genealogical relationships between different groups and dwellings in a village or several villages, then the methods in Chapter 6 will probably serve as well, or better, since genealogical diagrams of too great a scale tend to obscure relationships rather than bring these out. Certainly, as with any tools you intend to take to the field, you should learn and evaluate these well before depending upon them in a 'live' situation

3.3    Other types of data produced during fieldwork

In addition to the written data in §3.2, there are other common kinds of field data:

1.    Photographs, cine film, video
2.    Audio recordings
3.    Material culture; artefacts, measurements

These are briefly discussed in this section, and most have more extensive coverage in Chapter 5. What most of these have in common is that, for most anthropologists, these are at present somewhat peripheral activities, if well established (Blacking 1984:199) . Even in the case of photographs and to some extent audio recordings, which most anthropologists make, these are under-used for research, either in the field or out. On the other hand, most anthropologists I have spoken to are interested in using these materials, but have found them cumbersome in practice, so the material mainly sits in drawers.

Computers are available with the power, peripherals and software to address information of all sorts, not simply numbers, words, formulae and coded texts. This approach to computing is just beginning in the sense of being practical for ordinary users. In anthropology there is but one completed work which incorporates computer accessible images and ethnographic texts (Macfarlane 1989) as an ethnographic report, but there are a number of research projects in progress and proposed research under consideration. These initial projects are drawn mostly from researchers with a pre-existing interest in visual anthropology or folklore, but I believe that the application of computer technology is ultimately likely to radically change both the practise of fieldwork, ethnographic research, and ethnographic reporting. 

3.3.1   Visual recordings: Photographs, Cine film and video recordings

Anthropologists incorporate images into their research to a widely varying degree. Some take a few photographs simply for the purpose of illustrating some aspects of their notes and other data. Others use visual records as a central part of their research. Those in the former category may find little use for a computer other than perhaps the creation of a special register or card index of photographs and legends, and giving their photographs names and referring to these in the notes or other data resources. 

If visual records are an important aspect of your research, there are many more possibilities.The technical details of when, where and how to acquire images in the form of photographs or video, and the basic methods of analysis and use are no different from more conventional media for representation. These issues are well covered in Collier and Collier_2 (1986) and Jackson_2 (1967), as well as briefly discussed by Blacking (1984). However the range of analysis possible and the ease of access are more extensive. These range from text/code-based computer-assisted classification using authority lists (taxonomic thesaurus of classification terms for subject domains) to cross-references between fieldnotes and people, places and events depicted in the records, to the incorporation of the images into a computer representation which can be not only accessed and viewed on screen from databases, but can serve as a interactive element for data entry relevant to the images and objects in the images. Images can be incorporated into most computer applications which operate on graphical operating systems, including word processors, database programs. Images can be resized and otherwise manipulated using a paint program capable of editing colour images. 

Although it is well established technology, at present the capability to digitise and incorporate digital images comes at a price, in both equipment and weight (and money, of course). The computers are not a problem, since any notebook computer with greyscale or colour display can display images with adequate detail for most field use. There are, however, two basic problems with digital still images. Firstly, a method of acquiring and digitising images. There are a number of devices for digitising images, almost all of which are larger than one would like for fieldwork. The best all around field solution, if you do not require the best possible quality, is probably to use a video camcorder, with an attachment for slides and negatives, if required. This camera, in conjunction with a video frame-grabber peripheral will enable you to digitise stills directly from the camcorder. The frame grabber should come supplied with software for this purpose. At present there are no frame-grabbing devices intended for portable use, but most can be coaxed to run off battery by a competent technician. I expect this to be rather less of a problem in the very near future, with this capability built into to more expensive notebook computers. You are also likely to require more memory than you would require otherwise. With images of the resolution discussed eight megabytes of memory is probably the minimum required.

The advantage of this arrangement is mainly in terms of portability and time investment in the field. Hi-8 or SVHS-C camcorders can be quite small and light, as little as 900 grams+batteries and easily pocketable. The resulting image from a digitised frame will fill most computer screens with a full-colour (or greyscale) image rather better in definition than a good home-television. You can depend entirely on the camcorder for visual documentation (and sound as well), if this quality is adequate (it won't be any better once you leave the field). You can also take conventional photographs, and once developed these can be transferred to the camcorder using a slide-scanning attachment available from all major camcorder manufacturers. This leaves you the option of making better quality stills once out of the field.

The second problem is storing the images on a disk. Even highly compressed, a full-screen image will occupy about twenty to one hundred thousand bytes of memory (the equivalent of about five to twenty thousand words, rather than the proverbial one). This amounts to from twenty to fifty images per megabyte. If you produce about two thousand images in a short season of fieldwork, this requires about one to two hundred megabytes for disk storage. Given you have other storage requirements, this is rather a lot, but is an available option for most full-featured notebook computers.

For video tape, support for partial or even full transcription at frame level is available, either through computer-controlled players, or through direct representation on the computer. Indeed, full transcription may not be necessary in some cases because under computer control specific video frames or sequences can be displayed on the monitor as a response to database queries, along with notes and markings added by the researcher.

Because of current costs and limitations, using digitised video in the field has to be viewed in terms of application. For most researchers the principal application of displaying or making video in the field include: 

1      An interviewing aid. This suggests that you have to plan your requirements based on a system which you can either get your  consultants to, or get to the consultants. This may require only the capacity to display video if you can prepare the materials in advance. Software required can be as simple as a wordprocessor, although special purpose applications exist for this purpose (§5.3).
2      Interrum analysis of actitvites with a visual element. This will require facilities for producing video, although not necessarily at the highest quality.The kinds of analysis can range from frame by frame comparisons and documentation for interactional studies, to creating video clips with very coarse temporal quality; rates as low as one or even one-half video frame per second can give a good overview of the structure of an event such as a ritual or other ceremony.
3  Data acquisition. This will require facilities for producing video clips, though not necessarily of high quality. For example, with a temporal quality rate of one frame per minute or so you can record activity at a location for a day and analyse patterns of space use, traffic density or other periodic activity. 
4  Indexing video tape. This will require facilities for producing video clips, though not of high temporal quality. A index consisting of one frame per ten seconds for thirty hours of tape results in a computer document about forty megabytes in size. Time codes can be associated with the frames, which makes it very easy to locate sections of tape which you can watch using more conventional video equipment, including video decks which can be directly controlled by the computer. 

In all cases video clips can be manipulated by other computer tools, such as database management systems (§2.3) or inserted into wordprocessing documents or hypermedia authoring programs (§2.3.4), as well as specialised programs for working with video material .With the system level support indicated below, there is support in each of these applications for taking smaller clips from the original clips, regardless of what application they are embedded within, and installing these new sub-clips into the same or other application (Figure 3.6). This is done by reference, rather than making a copy, so the new subclip does not add appreciably to the storage requirements. A five minute video sequence can thus be broken into smaller and smaller sequences within an application such as a word processor, where these can be documented, while retaining the entire clip for reference. I have developed a prototype program for managing documents of this sort, which also alleviates some problems I have had with discussing visual materials with consultants in the field, which is keeping adequate track of what image or part of an image a segment of the refers to. This applications, currently called MovieSheet (Figure 3.6), can digitally record the interview into the visual document while tracking an on-screen pointer, that I position over the image or movie clip (and its temporal location) being discussed. At a later time the interview can be played back, including the movement of the on-screen pointer to indicate what image was under discussion, or only the part of the interview pertaining to a specific image or clip can be retrieved by activating that image or clip.

It is much simpler to display video clips (or still images for that matter) than to produce them while in the field. Any mid-range computer, notebook or desktop, can display digitised video at some level using free or inexpensive system level software 5. It is important to have system level support for video, so that you can use video clips in your other programs which are not designed to support  this kind of data. You can paste clips into wordprocessor documents and databases, as well as programs designed to use video directly .

You will get very poor quality from a monochrome screen, and surprisingly good results from a 16 level greyscale screen, although 256 colours or greys are much better. For notebook computers with flat screens, it is much better to have an active matrix screen rather than a passive matrix screen 6, although the former adds a lot to the cost. Some computers can output to a conventional portable colour television, which is usually inadequate for text but displays digital video in a plausible manner.

The problems relating to producing digital video in the field at present are similar to still images, only compounded - requirements for more special hardware, memory and much more disk storage. Although the use of digitised video in the field is one of my current research interests, I can recommend this only if you are willing to make a large investment in time and money, as at present this requires a lot of equipment and expense. Although I will summarise current hardware considerations, you should consult a support person about the latest  developments, as the terrain is changing almost daily at the moment. 

Presently twenty megabytes of RAM memory is fairly comfortable for video work involving small image size (about  one quarter of a normal screen,about 320 pixels by 240 pixels). This memory is not required to play back the video sequences, only to produce them. Any amount of memory up to one gigabyte (230 = one gigabyte = 1,073,741,824 bytes)would be very welcome. 

In terms of disk storage, one minute of quarter-screen size video at 15 frames per second requires about four megabytes of storage. Full screen video (640x480) requires almost four times as much. Currently, this factor alone currently vastly reduces the portability of a system for working with video, although it can be overcome. 

If you want to take video documents to consultants, but can produce video at a base, you might consider two different computers. With current software support  digitised video at 10 to 12 frames per second (fps) is easy and acceptable, if crude, for ethnographic work, and 15 fps is possible with faster notebook computers. Note that video at both these rates will be rather crude, but is good enough to be useful. The standards for conventional video are 25 fps for PAL and SECAM, and 30 fps for NTSC video standards. It is possible to attain these speeds, even at full screen resolution with expensive hardware support, but  in my opinion is not at present a reasonable option in the field unless there exceptional reasons.

For further discussion of digital video see §5.3..

3.3.2   Audio recordings

The range of options for audio recordings fall under similar lines as visual recordings (§3.3.1). For the serious user of audio tape similar options for computer representation and control are available. Computer controlled audio decks can locate and play tape sequences on demand from either direct requests by the user, or in as a result of searching a database or from a fieldnote reference. Audio of reasonable to high quality can be stored directly on disk, and a number of programs for editing and modification are available. 

Although the options are similar, digital audio recording directly to the computer poses few problems other than disk storage, and this is not of the scale of digital video. One minute of audio suitable for voice or low quality music reproduction can be stored in about one-half megabyte of disk (with some compression). At compact disc quality one minute is about ten megabytes. The equipment necessary for low-to-medium fidelity sound is quite inexpensive and trivial in weight, size and power requirements.

Like images and video, sounds can be imported into word processing documents, databases and other computer tools.Audio clips can be manipulated by other computer tools, such as database management systems (§2.3) or inserted into wordprocessing documents or hypermedia authoring programs (§2.3.4), as well as specialised programs for working with audio material.With the system level support, there is support in each of these applications for taking smaller clips from the original sound clips, regardless of what application they are embedded within, and installing these new sub-clips into the same or other application. This is done by reference, rather than making a copy, so the new subclip does not add appreciably to the storage requirements. A twenty minute audio sequence can thus be broken into smaller and smaller sequences within an application such as a word processor, where these can be documented, while retaining the entire clip for reference. 

Like video, in some cases there is no need to fully transcribe audio material in this form, since direct and immediate access to the audio clips can be directed from a computer-based document, while retaining the context of the clip.

3.3.3   Material culture, artefacts, physical measurements

Standard computer applications can greatly ease the difficulty of cataloguing, describing and referencing instances of material culture, including photographs, artefacts, botanical and biological specimens. The use of authority lists (recommended for manual management ) simplifies the problem of systematic treatment, as well as access to the references or even images of the materials (§3.3.1). As with other data, active cross-references can be linked to other record types, which also improves access.

Although measurements of length, distance, temperature, humidity etc. are usually transcribed from instruments to paper (or a computer), if you use a lot of measurements of a particular sort, it may be advantageous to log measurements directly into a computer database using a special version of the instrument which hooks directly to a notebook (or smaller) computer. Otherwise guidelines for special registers are adequate (§3.2.6).

3.4   Ancillary Field Activities

There are a number of activities in the field, which while not research, contribute to the process of research. Although you would never take a computer into the field for these purposes, they are worth considering once you have done so.

These generally cover basic functions, such as:

a)    Vocabulary drills. The word lists you develop can easily become electronic flash cards.
b)    Field accounting for expenses, which can make good use of either database programs or spreadsheet calculators, especially handily when using multiple currencies (Ellen 1984).
c)    Making appointment files, for which special purpose programs are available.
d)    Making official-looking correspondence, which has been of considerable benefit to me on several occasions.
e)    Occasional entertainment, if you fancy a game of chess from time to time. Video games should be avoided by those weak of will.