Tables 2 to 4 above show that eye movement was commented on more often than head or hand, suggesting that there was a lot of eye contact between the actors – as might be expected from active interaction in general. This suggestion is also supported by the large number of nods as well as blinks, which usually accompany conversational events. Most of the time, speakers kept their hands flat and again alluded to the non-confrontational nature of the dialogues; while the predominant use of the right hand against the left hand speaks of the dominant rightness of the actors. Even though our corpus structure followed the main features of multimedia corpus annotation in general, in particular the DiAMSL framework (cf. Bach and Robert, 1979; Sperber and Wilson, 1986; Wilson and Sperber, 2002; Bunt et al., 2010), differed from them in that in addition to assigning some annotation labels following a multimodal observation, we also relied on the unimodal observation. While multimodal observation involved all available modalities (usually both video and audio), unimodal observation only followed the video or audio of the given recoding. The reason for this additional unimodal observation was that we wanted to better identify which of the modalities in question has or has a specific contribution to the performance of certain communicative/pragmatic functions, including intentions and emotions. In this way, we hoped for a better understanding and a more detailed description of the individual differences between speakers in the expression of these functions as they are perceived and interpreted by observers. The labeling of verbal and, in particular, non-verbal events for behavioral categories followed a sophisticated protocol in which annotators were continuously involved in discussions. The specificity of the differences between the subjects was captured through a significant period of familiarity with each of the subjects` overall behavior on the recording before the actual annotation could begin.
Here is the example of very short dialogues consisting of disagreement of agreement and disagreement in the dialogue: this diagram tells us that the speaker turned his head to the right twice (“v_head” = movement of the head commented with video observation, “b” = beginning, “e” = end, “right” = movement of the head to the right). The addiction characterized by parentheses also showed that the first movement of the head was shorter than the second. Overall, the occurrence of “beginner” events in models was about 16% higher than that of “final” events, showing that the beginning of an action was somewhat more specific to a model than to its end. Figure 1. Real vs random (top) and random deviation (below) using file: i082. Agreement is not an autonomous state of mind of an individual: it is a behavioral event that necessarily involves an interaction that requires at least two actors and a subject. It arises as a reflection on the veracity of a statement, a point of view or an opinion and can develop under at least two conditions: (a) during the interaction, the actors realize that they independently share the same point of view, or (b) one or more of the actors are convinced by the reasoning of the other actor(s). The agreement process takes different forms depending on these two different conditions: if actors A and B independently share the same point of view, actor B`s agreement usually follows a statement or elaboration of actor A as a feedback channel of one kind or another (as yes, indeed!). If actor B is convinced by actor A of the veracity of a certain point of view, actor B`s act of consent can follow a question or request from actor A (e.B.
What do you think? or do you agree?), but also other scenarios (e.B. those with non-verbal events or pauses, practically anything that leads to a turn change) are possible. Like the agreement, disagreement also develops in response to an earlier request (Kakavá, 1993; Locher, 2004). . . .
Comments are closed.