Certain characteristics make a tape more or less adequate for Interaction Analysis. Ideally, one would want the camera to capture all relevant aspects of the event as it unfolds. But since it is impossible to know ahead of time what, in the course of analysis, will emerge as relevant, compromises must be made for most tapings between what is desirable and what is possible.

The Ethnographic Context A very important consideration for us is the extent of prior ethnographic fieldwork. We feel that ethnographic work is crucial not only for proper contextualization of tapes during analysis, but also for negotiating consent, setting up a relationship with participants within which they are comfortable with taping, gaining their cooperation for review sessions, scouting out the best camera locations, making decisions about pipping (see below), and thinking about what will not be visible on tape. As a matter of fact, it is probably desirable to have ethnographic work continue to whatever extent possible during videotaping.

We have found writing concurrent fieldnotes while taping extremely useful. As Minneman (1991) comments: "Notes taken during the event help in any subsequent analysis. It is immeasurably easier to jot down the time at which the group was doing something interesting for closer examination than it is to find that instant in a pile of two-hour videotapes later" (p. 98). Beyond that, such notes may allow one to explain otherwise inexplicable occurrences and avoid false interpretations. In one videotape of a birth, for example, the midwife raises her head every so often, ostensibly looking at the husband of the woman in labor. In fact, she was looking at the clock on the wall behind the husband. Knowing this fact (which was available from fieldnotes, but not from the tape) produces quite a different reading of the significance and timing of those looks.

A sketch of the scene indicating camera position, location of salient objects and technologies, features of the facilities such as windows and doors, and above all, people who are not visible on the tape also aids substantially in arriving at an understanding of what goes on. Ethnographic work may also include collecting crucial artifacts (for example blocks used in design problems) and making copies of important documents such as minutes, design and engineering drawings, math problem solutions, writing exercises, and discards of various sorts, all of which may become important for the interpretation of the tape.

The Physical Setup At the risk of sounding elementary, we offer some oft-neglected cautions. Set up the video and audio recording equipment on stable mounts (e.g., tripods, tables, book cases) so that they will not fall over or have to be attended to during the activity. Make certain that the equipment is positioned so that it does not interfere with the activity. In particular, ensure that participants will not trip over power and interconnection cabling. Do not place the camera so that it points into an open window or other strong light source because it will white out the picture. Finally, if the activity will last longer than one recording tape, make sure that you have access to the recorder without interfering with the activity. If possible, set a timer that will remind you that the tape is about to run out.

Shooting Social Scenes We make every effort to have all participants in the picture and to have their whole bodies visible. This generally means using a wide-angle lens. A common temptation is to point the camera at the person who constitutes the center of attention, such as the teacher in a classroom or the surgeon performing an operation. In the extreme, this leads to the phenomenon of "talking heads," a format familiar from television interviews where the camera flips from speaker to speaker, excluding not only other persons in the scene but also the rest of the speaker's body. Such tapes do not provide good data for Interaction Analysis, which is based on the premise that what the speaker says or does is fundamentally a social phenomenon, orchestrated with, and responsive to, other individuals in the scene.

Multiple Cameras In unstaged group interaction, people are generally not orienting in the same direction. Therefore, no matter where one puts the camera, some faces will be turned away. One partial solution to this problem is to use multiple cameras. Two cameras also are effective when the interaction involves artifacts or spaces whose state has to be tracked. For example, one camera can be kept on a computer screen or design work space, while the other records people's activities.

Roving and Stationary Cameras Experience has shown that, if the action is reasonably predictably confined to one particular place, it is best to keep the camera stationary so that a consistent and sustained view of the scene of interaction is obtained. This is preferable to a roving camera, which catches short stretches here and there, but typically does not get the all-important transitions. In addition, a roving camera is maximally intrusive. Much of that intrusiveness is removed if there is no operator behind the camera with whom participants expect to make eye contact. On the other hand, in situations where participants are thoroughly accustomed to a camera and where the ethnographer knows exactly what he or she is after, a roving camera may be the best method for getting maximum information. However, it is important to keep in mind that such a record reflects to a great degree the momentary judgments of the ethnographer. These, in our experience, are often questionable in retrospect.

Pipping PIP (picture-in-picture) is a useful procedure which allows two or more simultaneously recorded images to be merged and recorded onto the same tape, either at the time of taping (thereby saving on videotapes) or later. The secondary images typically appear as small insets in a less important region of the primary image.

However useful it may be to have multiple views appear side-by-side when doing analysis, pipping a tape also has its costs. The pipped-in picture overlays some piece of the primary tape, losing that piece to viewing. It is often possible to locate the pip in an "unimportant" area of the screen but that requires constant monitoring and moving the pip. In addition, when viewing a pipped tape, the second (or third, or nth) picture is often quite distracting. A screen with several pips is difficult to parse even for experienced analysts and is a liability in review sessions with inexperienced participants. If the pip is done in the field (which means that it cannot be removed later) it should be turned on and left on only for very good reasons. It should never constitute the default.

When selecting camera positions for shooting tape to be pipped, it is well to remember the film makers' rule never to cut to a view that is 180 degrees reversed from the preceding shot because the viewer can't make the transition. We are often tempted to put diametrically opposed views together, such as one shot from the back of a room with another from the front. While such arrangements largely solve the problem of covering the scene adequately, they make the resultant compound image confusing and difficult to parse, especially for participants but even for experienced analysts.

When, then, is a pip useful? We have often found that we need a pip when people are engaged in monitor work. For the analysis of such situations one needs a simultaneous view of how the computer responds to what operators do and how that, in turn, enters into their interactions. It is also useful for situations where one wants to focus on a specific activity and the background for that activity at the same time. For example, one might pip the picture of a teacher and her or his activities in front of the classroom into a whole-classroom shot. By contrast, such times as transition between one class period and the next, when a crowd disassembles and then reassembles, are probably best shot with a single wide-angle camera. In general, if the phenomenon of interest consists of people coming in and assorting themselves in interactional spaces, it is almost always better to use a single camera with a wide-angle lens.

The Importance of Sound What is often more consequential than multiple cameras is high-quality microphones. We have found wireless mikes particularly useful because they allow mobility. We have also found that placing multiple tape recorders into the scene aids substantially in later disambiguating what people in noisy multi-actor social situations say. Sometimes one may be able to ask a roving teacher or supervisor to carry a tape recorder. Some investigators have made up special packs that can be carried even by small children tearing around in a playground, where their shouts and laughter could never be adequately covered by a stationary mike even when a stationary wide-angle camera covers nonvocal activities adequately.

As with any technology, knowing what the strengths and limitations of the recording equipment are is essential for making tapes that are suitable for studying complex social interactions. In the final analysis, a tape will be as good as the amount of forethought, practice, and careful preparation that has gone into it.