Coordinating Meaning Through Electronic Messaging: A Lexical Analysis of Healer/Patient Communication

Theoretical Background and research questions/hypothesis: This study combines Natural Language Programming and Lexical Analysis of Data with Pearce and Cronen’s (1998) notion of Coordinated Management of Meaning (CMM) in an effort to determine whether or not the extent of shared meaning achieved over time can be measured in electronic messaging between physicians (“healers”) and patients. The initial goal was to describe the nature of physician/patient electronic communication and then to search for emerging trends within the data set.  First the data was de-identified.  Second dictionaries were developed to be used as word searches.  Third, a computer program for flattening and coalescing the electronic message threads was created. Eventually, several hypotheses emerged: The longer physician/patient dyads communicated with each other over time, the more meaning was shared.  The more transactions and dyads overall that physicians participated in, the more the dyad adapted their language patterns to each other (CMM). Those physicians who interacted in more transactions and dyads also adapted their use of Medical Jargon over time to match with the patient’s use of the same jargon.

Methods:  Roughly 60,000 electronic messages between 54 physicians and 25,000 patients using HealthTrak secured portal within Montefiore Hospital’s Internal Medicine Program of the University of Pittsburgh Medical Center (January 2006-April 2010) were de-identified and analyzed.  First each unique word was coded into 30 interrelated “dictionaries” creating a tautology of categories unique to medical communication.  Second, the messages themselves were converted into Type/Token Ratios (TTR).  Various statistical tests compared dyads’ serial and cumulative TTRs and overall usage patterns for high, moderate, and low physician users.  Dictionary categories were also converted to TTR scores for each message.  Outliers were determined and comparatively analyzed to see to what extent each differed in TTR scoring.

Results:  All hypotheses from the initial exploratory study were supported. An inverse relationship between physician number of dyads/transactions and TTR scores was found which suggests that physicians and patients mutually adapt to each other’s language patterns through a coordinated management of meaning.  This was particularly true for dictionaries such as “Medical Jargon,” “Emotion,” and “Courtesy Terms.”  That is, as interactions increased, TTR scores for individual categories of words decreased.

Conclusions:  Although it is impossible to measure if the interpretation of messages between physicians and patients are identical, it is possible to watch how the sharing of words (especially jargon terms) over time occurs, assuming that as mutual repetition of words increases, shared meaning also increases.  Despite the argument that electronic messaging may not be as effective as face-to-face (FTF) communication, this study reveals that language adaption within electronic messages develops over time in a very similar fashion to how it does in FTF encounters.

Implications for research and/or practice: A future NIH grant is being sought to further develop programming for lexical content comparisons between various groups of users as distinguished by culture, gender, and socioeconomic background.  Adaptive methods will be observed and hope to lead to a statistically supported training program for medical students and healthcare professionals alike.