«Andrés Gregor Zelman The University of Amsterdam 2002 ii Mediated Communication and the Evolving Science System: Mapping the Network Architecture of ...»
70 File Occurrence Unique Words Specificity Transmission 283992 7627 0.77 0.28 Print-All 226362 1009 0.81 0.08 Print-Shared Table 4.6: Print System Dynamics The measure of Specificity suggests that the distributions of word use in Print-All and Print-Shared are not so different, thereby indicating a reduction of uncertainty.16 Of the words in the Print-All dataset, 23 % are stable, compared with 19% stability of the Print-Shared words. This measure has also revealed that those words which have little effect on the specificity are not located in the Print-Shared dataset. The restriction in the number of words used during the two year time period proved more significant when compared with the shared mutual information across the time periods. While specificity refers to the specificity of the total word distribution, transmission eliminates the words not shared across the four time periods, and hence there is a reduction in the mutual information shared. The analysis revealed much less transmission in the Print-Shared dataset than the Print-All. The transmission of words in the Print-All dataset is three and a half times better than in the Print-Shared. The higher transmission in P-All is because of the inclusion of words which do not occur in each time period; thus, words which were not shared carried the transmission.
Generally speaking the expectations of the systems analysis were not confirmed – no evidence of self-organized criticality was found. However, the additional analyses of the levels of specificity and transmission in the print dataset showed that the distributions between the Print-All and Print-Shared shared a similar stability as regards specificity, and that the transmission of words in the Print-All document was significantly higher than the Print-Shared.
The primary objective of the architecture analysis was to identify whether the SOEIS print communications could be shown to have a discernable architecture. If so, it was expected that this architecture would reveal evidence of codification of scientific information. Confirming our expectations, the distribution of the standardized mean ratio percentage of unique words compared across the datasets revealed a stable process of information codification. Indeed, evidence was found of both an a priori codification and a processual codification. This line of argumentation was contextualized using the Medium Theory notion of the Information Network. The secondary objective here was to determine if particular qualities could be identified with a decidedly print mode of communication. Arguably the analysis has revealed architectural qualities of the print dataset, but only in juxtaposition with the results of the analysis of electronic communication can this expectation be confirmed.
The central objective of the network analysis was to identify network properties of the SOEIS print communication by comparing the fluctuations of keyword use over the 16 The measure of Specificity indicates the way that the information flow is specific in deviating from 100%. The specificity of the communication in terms of collective word use in the case of Print-All is 77% specific, and in the case of Print-Shared, 81 % specific.
71 four time periods of the analysis. It was expected that a discernable pattern of keyword reoccurrence would be visible: some words would remain constant while others would likely occur in specific time periods, thereby showing particular emphases and evidences of information codification. These expectations were confirmed using a range of analyses. First, by comparing the top 50 keywords a general sense of the cognitive orientation of the research project was revealed. In order to gain a deeper sense of the topics that were addressed in this mode of communication, the document sets were compared with the total document set, and then against each other. These two latter approaches revealed different, and arguably more salient, sets of central topics. Substance was thereby added to the architectural frame by revealing how networks between co-words and their collocates differed over the time dimension. Actor Network Theory provided a focal point for understanding the differences in keyword use to reflect cognitive developments. Here the symbolic dimension of the SOEIS print communication is relevant and is best interpreted in light of the networks of intertextuality outlined in Chapter II: Theoretical Grounding.
The structural network of relations isolated by Saussure’s notion of difference can be seen here to reflect the closed nature of the print dataset. Yet the dataset is not an isolated language system, it was created over time and accumulated; it was precisely the collected traces of the SOIES’ use of the print medium. Hence the closed nature of Saussure’s network of associations doesn’t capture the essence of the print dataset. It is perhaps best conceptualized with respect to Derrida’s notion of différance which highlights the inter-textual nature of the individual words and their interrelation. The print communications network of relations was contingent upon a social context, and therefore when interpreting the frequency or disappearance of keyword occurrence it proves more beneficial to appeal to the Postructuralist notions of meaning and exchange. Structuration Theory provides one important additional perspective to this theorizing about measurable meaningful exchange. The dynamic process of wordreuse can be interpreted to reflect properties of social and system integration; the social is integrated by the SOEIS members using the print medium – the structure of the language system is thereby operationalized – and the communication system writ large is also (re)integrated (which by definition remains unobservable).
Finally, the third analysis addressed the question of whether path dependencies in the SOEIS dataset could be identified, thereby indicating systemically that critical transitions were necessary for the communication to develop. Here it was expected that when examined for path dependency, points of critical transition would be discernable, thereby indicating that each stage of the project was necessary given the expected information content. This expectation was not confirmed. However the supplementary analysis performed did reveal an unexpected quality of the print communications – that the transmission of words over the Print-All dataset was significantly higher than that across the Print-Shared dataset. This final observation suggests that, in light of the macro perspective offered by Self-Organization Theory, despite the inability to find critical transitions or path dependencies in the dataset, the measures of specificity and transmission reveal a particular codification of information. This result may prove important when compared with the results of the system oriented analysis of the electronic communications dataset in the following chapter.
72 This chapter has served to characterize the SOEIS print communications with respect to its the architectural, network and systemic dimensions. In so doing the modelling aspect of mapping mediated communications was addressed, as were the symbolic, or meaning dimensions of the communication. Further, this approach has established a mode of analysis that will enable a comparison between the print communications and the electronic. Importantly, the Architecture – Network – System triad is central to binding this analysis. It is treated here as the core model of this dissertation, and will be repeated throughout Part II – Analysis. It will prove crucial to collating and comparing the results of all analyses in Part III – Reflection.
This chapter subjects the electronic output of the Self-Organization of the European Information Society (SOEIS) research project to the same analyses as the print communication analysis described in the previous chapter. Again, the examination entailed a threefold analysis to explore the architectural, network, and systemic dimensions of the SOEIS communications. A computer assisted textual analysis was performed upon the body of the SOEIS electronic communications, as recorded in the email exchanged on the two project mailing lists (EuroCon-Knowflow, and SOIS).
Several different techniques are employed to understand the characteristics of word
use in the mailing lists for the project time period, as described in Chapter III:
Materials & Methodologies. By using precisely the same techniques for the print and electronic analyses, a method of comparing print and electronic output is established in order to understand the roles that different media play in processes of knowledge production. In this way it proves possible to assess the degree to which print and electronic media bias different types of information organization; or, indeed, if perceived as processes, of knowledge production.
Importantly, like print, electronic writing leaves examinable traces and it is precisely these traces and their interrelation that are examined in order to uncover a network structure in the data. Similar to the print analysis, the primary aim here was to use textual analysis techniques to determine whether the SOEIS electronic communications could also be shown to exhibit architectural, network, and systemic properties. In this way the central question was addressed: can electronic knowledge production as a mediated process be shown to exhibit a particular media bias? This led to the question of whether the network architectures of print and electronic media could be compared to reveal particular biases implicit in mediated processes of knowledge production. The second aim of this research was therefore to compare the results of the textual analysis of the electronic dimension with the results of the analysis of the print dimension. Using this approach, the relevant similarities and differences between these respective modes of communication become apparent.
The analysis of the electronic dimension first sought a structural architecture to the data. Networks were then located through analyzing keyword fluctuations, thereby illustrating electronic communication as process of mediation distinct from print.
Finally, the electronic dataset was examined for self-organized criticality and the results compared with the results of the system oriented analysis performed in the previous chapter. Media is emphasized here as something used – as described in Chapter II: Theoretical Grounding with the aid of Structural, Postructural and Structurational theories of network and meaning.
The central hermeneutic unit remains the same in this analysis: the keyword in context serves as a focal point for questions concerning both individual and collective 75 behaviour. Again, keywords and associates were treated as events that compile and complexify via networked communication and that may prove to behave systemically.
Importantly, language used in email communications is expected to be fundamentally different from that used in print communication. This may be explained in part by the formal nature of print communications and publications as distinct from the ways that
language use operates in email communications, as discussed in Chapter III:
Materials & Methodologies. Notably, the SOEIS email communication is fundamentally different in its consistency than the print communication simply because it is collectively generated as individual messages, whereas the print communications were mainly co-authored communications.
The electronic output of the SOEIS members was examined for its basic architectural features including the document size, word frequency and unique-word frequency (and their ratio), for network features such as rate of word-use change, and for systemic features such as phases of ‘pathway dependence’ or critical revisions / transitions in the dataset, and then differences in overall word distribution. The differences in the consistency across each of the four respective electronic documents are relevant here in how they compare with the consistency across the print documents.
Research Questions & Expectations
Given that similar information was exchanged over these two different information channels it was expected that differences between the respective databases could prove instrumental in finding particular media bias. Further, given the centrality of the print medium to the functioning of the SOEIS research project, the analysis aimed to determine if print provided a decidedly different mode of communication than electronic media and if this difference had any notable impact on SOEIS communications. As with Chapter IV: Analysis of Print Communications the basic expectation here was that an analysis of the architectural, network, and systemic features of electronic mediation will help to reveal biases particular to this medium.
The theoretical triad of Architecture – Network – System framed this analysis, provided a basis for comparison, and assisted in the formulation of a number of distinct expectations. Three research questions were thereby identified, corresponding with the theoretical lens. The research questions are described below with the central expectations of each analysis.
First, does SOEIS electronic communication have distinctly different architecture than the SOEIS print communication, and can particular qualities be identified with a decidedly electronic mode of communication? As with the examination of SOEIS print communication in the previous chapter, registered changes in the mean ratio percentage of unique words in the electronic communications were used to identify cognitive orientation, and to isolate processes of information codification. In this way, a network architecture unique to the SOEIS electronic communications should be revealed. Given that the architecture analysis of the print communications in the previous chapter revealed evidence of information codification, the primary expectation here was that a comparison of the basic architectures of the print and 76 electronic databases would reveal qualities particular to each mode of communication.