The elicitation data must be transcribed to bring out the gesture and speech data. For details, download the annotative practice .pdf, which summarizes the process (and is more up-to-date that the appendix in Hand and Mind.
A gesture as it is usually defined passes through up to five phases: preparation, prestroke hold, the stroke itself, poststroke hold, and retraction; all are optional except for the stroke. The stroke caries the imagistic content of the gesture and is the phase whose synchrony with speech is maintained by the speaker. The following transcription illustrates many of these features (transcription by S. Duncan; the illustrations show three stages of the gesture):
/ tryi[ng to swing across by a rope #]
prep hold stroke hold retract
Transcription: Iconic; 2 similar hands; A-shape; palms
body; fingers turned down;
starts at right and arcs to other side with slight wrist pivot. Hands
=S's hands, character vpt = S; arc = trajectory, observer vpt. S swings
A misconception has arisen about the nature of the gesture categories described in Hand and Mind, to wit, that they are mutually exclusive bins into which gestures should be dumped. In fact, pretty much any gesture is going to involve more than one category. Take a classic upward path gesture of the sort that many speakers produce when they describe the event of the cat climbing up the pipe in our cartoon stimulus. This gesture involves an iconic path-for-path mapping, but is also deictic, in that the gesture is made with respect to an origo --that is, it is situated within a deictic field. Even "simple" beats are often made in a particular location which the speaker has given further structure (e.g. by setting up an entity there and repeatedly referring to it in that spatial location). Metaphoric gestures are de facto iconic gestures, given that metaphor entails iconicity. The notion of a type, therefore, should be considered as a continuum --with a given gesture having more or less iconicity, metaphoricity, etc.