This thesis investigated the link between trust and language in online dyadic, online group, and virtual community interactions. The first two empirical studies revealed that positive emotion is reliably linked to trust, that is to say that dyads and groups that produce more positive emotion words show greater trust. In addition, by priming group members to believe their group was either high or low in trustworthiness, the second empirical study revealed that group members produce more positive emotion words as a consequence of greater trust.
A third study revealed an additional important language phenomena to trust, this was a type of linguistic mimicry known as linguistic style matching (LSM). The study guided groups through established phases of group development (relational, task, and task resolution), findings revealed that LSM in the early, relational phase, predicted trust. Further analysis of the relationship between phase of development, positive emotion words and LSM revealed that when LSM was low in the relational phase, positive emotion once again became a cue used to trust in subsequent task phases.
The Relevant and Reliable Language Theory was proposed to explain the phase of group development dependent effects of LSM and positive emotion on trust. In short, the theory states that language must be relevant, i.e. the group is still undecided regarding trust judgements and still seeks trust relevant information, and reliable, i.e. in a given context, the language must be considered a genuine trust signal, and not one that appears ‘faked’ by ‘opportunists’. Based on the novel theory, it was hypothesised in a forth study that disrupting LSM in an early relational phase, and positive emotion in a subsequent task phase (contexts where LSM and positive emotion were proposed to be most relevant and reliable to trust), would have the most detrimental effect on trust. A confederate implementing these disruptions, relative to a comparable control, supported the theory; with groups who were subject to the ‘strong’ disruption producing lower trust behaviour than the control group.
A final study extended finding to real world online interactions in a virtual community focused on discussing credit card fraud, i.e. a criminal online group. The discussions were relational, rather than task focused, thus The Relevant and Reliable Language Theory predicted that LSM would be the most important language variable related to trust. The theory was supported as LSM, but not positive emotion, ebbed and flowed in line with predicted levels of trust.