«Andrés Gregor Zelman The University of Amsterdam 2002 ii Mediated Communication and the Evolving Science System: Mapping the Network Architecture of ...»
8 As argued in Chapter IV: Analysis of Print Communication, since the time periods do not contain the same amount of words shared between the time periods the calculation is presented with a division by zero; this is problematic because the word then doesn’t occur as a predictor (one cannot divide by zero). The solution to this problematic was found by comparing only the shared words between the four time periods.
87 Between the first and second periods of the Electronic-All dataset,.184 bits of information was shared and between the second and third time periods.378 bits of information was shared. When compared with the bits of information shared nonlinearly between period one and period three (.303) it was learned that no critical revision of the texts occurred between periods one and three via period two because the total number of bits was higher than the non-linear pathway. When all time periods were compared using this approach, it was learned that there were no critical transitions in the full reference corpus of 7672 words. And again, similar to the PrintShared analysis in Chapter IV, when the Electronic-Shared document set (comprised of word lists based solely on the basis of their inclusion in each of the four time periods) was examined, no path dependencies were identified. These results are presented below in Figure 5.7: Four Time Periods Compared Linearly (1759 words)
and Figure 5.8: Four Time Periods Compared Non-Linearly (1759 words):
With this analysis we find that there is no critical revision of the information at the level of the email communication, in either the Electronic-All or Electronic-Shared datasets. Thus, neither the electronic nor the print had any periods in which the overall distribution of words being communicated over the two year time period of the SOEIS project significantly changed. However, it is notable that in the Electronic-All dataset the transition between the third and fourth time periods is significantly lower than evident in the print dataset. This fluctuation in the electronic dataset may be attributable to the newness evident in email communications, or may in fact be related to the notable dip in the size, word count and unique word occurrence isolated in the architecture analysis. The lack of fluctuation in the print case reflected the codified nature of print project communications as described in Chapter IV: Analysis of Print Communication, which suggested a lack of innovation in heavily codified EU projects.
The final analysis of the electronic database examined the four respective wordlists for characteristics of word distribution. Since there was no critical transition apparent in the electronic communications we turn again to the architectural parameters of the 88 dataset to identify changes in Specificity and Transmission. Again, Specificity is the specificity of total word distribution across the whole dataset, as the ratio of the expected information content of the distribution relative to the maximum information content – it is the measure of the degree to which some words occur differently in the electronic dataset along the time dimension. Transmission, as the flow of mutual information across the dataset, represents a reduction of this specificity, or uncertainty – it eliminates the words not shared across the four time periods. Table 5.6: Print / Electronic System Dynamics shows the overall word Occurrence, Unique Words, Specificity, and Transmission of the print and electronic datasets.9
The results were compared with the results of the print analysis. With respect to unique word count there was a notable difference found between the Electronic-All dataset and the Electronic-Shared dataset; the difference between the unique words in Print-Shared with Electronic-Shared indicates that print set performs the same task (the carrying of the SOEIS communications) with a smaller number of words than in the electronic set. This implies a restriction in the number of words used in the project’s print communications. The restriction in the number of words used over the course of the project is significant when compared for the Transmission of bits of information. There appears to be more transmission between the time dimensions in the electronic case, but the difference is mainly caused by words that occur in specific periods and not in others. Thus, there appears to be a stronger process of codification in the print documents than the electronic documents. This claim is supported by the observation that specificity remains lower in the electronic case than the print and that this observation does not change with the shared datasets (0.99). Compared with the specificity of words in the Print-All dataset (23 % stability) and the Print-Shared words (19% stability), the Electronic-All and Electronic-Shared both exhibit a mere 1% stability – the word distribution in the electronic dataset is nearly random. The print dataset is therefore much more structured and codified than the electronic.
The theoretical triad of Architecture – Network – System as the model through which the analyses were performed and interpreted has aided in the collation of the results obtained through the analyses of the SOEIS print and electronic communications. The main objective of this chapter was to examine the electronic communications using the same methods as the print analysis in the previous chapter, and to compare the results. Elements associated with print as a Mode I type of knowledge production and 9 The occurrence and unique word counts shown in Table 5.6: Print / Electronic System Dynamics differ from the occurrence and unique word counts found in the print and electronic architecture analyses (above) because the texts were filtered using the adapted stop-list.
It was asked if the electronic communication of the SOEIS would prove to have a different architecture than the SOEIS print communication, and if particular qualities could be identified with a decidedly electronic mode of communication. It was found that the electronic communication does have a distinct architecture from print. It was learned that email activity decreased over time, while size, word frequency, and unique word occurrence increased over the time period under analysis. The unique word percentage and mean ratio percentage of unique words dipped only slightly in the third period due to the shift of dominant member participation from the EuroConKnowflow mailing list to the SOIS mailing list during this period. The mean ratio percent showed that the style of variation of the electronic database increased over the time periods, thereby suggesting an influx of new terminology, ideas, and presumably new research interests. Thus, the print and electronic architectures proved to be quite different; while the print architecture exhibited a rather stable codified communication the electronic dataset appeared to be structured differently – the increasing variation over the time periods proved more expansive than constricting. It was expected and shown that the comparison between the print and electronic unique word percentage would reveal particular qualities of each mode of communication (print, electronic).
This was certainly the case, with the results of the electronic analysis suggesting a more Mode II oriented communication than the Mode I oriented print communications. The differences highlighted in the architecture analysis were interpreted in light of the Medium Theory notion of the Information Network; the differences were shown to be significant with respect to the mean ratio percentage of unique words.
The question of whether network properties of SOEIS electronic communication could be found by comparing the fluctuation of keyword use over the four time periods was then addressed. It was expected that this comparison would reveal clear differences between print and electronic modes of communication, given their respective functional roles in the SOEIS project. Using the same method as those used in the print network analysis, the top electronic keywords were found to contain similar project title words. By comparing each set of texts with the document set, and then set with each other, a number of additional conclusions were reached. The networks of electronic keywords were shown to supplement the activities of the research project, but those words associated with the functioning of the research project itself decreased in emphasis over the dataset while similar words in the print dataset were found to increase over time.
In contrast to the print dataset, where it was discovered that most words occurred in the final document set thereby suggesting an aggregative text, the word occurrence in the electronic dataset appeared more evenly distributed. Thus, while the print served an archival function where text was produced, reworked and resubmitted, the electronic served a more supplementary role whereby project issues per se were not discussed. The emphasis, rather, appeared to be on the use of the list as a means to communicate about the substance of the studies in the research project. The expectation of this analysis was confirmed; the networks of keyword distribution revealed herein have proven significantly different than the networks fostered via 90 print communication. This difference can be interpreted in light of the Poststructuralist, and Structurational networks of meaning and intertextuality.
Networks of association in language were argued by both Derrida and Giddens to be contingent upon language use in social context; this underscores the importance of this comparison of how different contexts demand different (word) production habits.
Print and electronic writing integrated the research group in different ways, but the communications were aimed toward a common goal. Structuration Theory is relevant here because the dynamic process of word-reuse, as cited in the last chapter, reflect the properties of social and system integration – importantly, this integration is different with print and electronic forms of expression. It is argued here that these two media served to integrate the SOEIS communications in measurably different ways.
The meaningful networks isolated in both print and electronic network analyses emphasized how each served to socially integrate the SOEIS system in decidedly different ways.
The third and final research question addressed in this chapter asked whether path dependencies or critical transitions could be identified in the SOEIS electronic dataset. If so, this would indicate that the information exchanged went through necessary revisions to become the final product. It was also asked if this would prove to differ from the results of the print dataset. No critical transitions were found in either the print or electronic databases. However, despite this apparent lack of systemic properties, differences were found through the comparison of the transmission and specificity of word use in each respective database. With respect to unique words count, the print database appeared to perform the same task with a smaller number of words than in the electronic set thereby suggesting a completion of the same task with less talk. The SOEIS print provided a more codified communication than electronic. When compared for transmission, or mutual information, the electronic database was shown to exhibit more transmission between the time periods than the print analysis, and this difference was mainly caused by words that appeared in specific periods and not in others. Thus, the print database has proven to be more codified than the electronic, and this observation was supported by the observation that specificity was shown to remain lower in the electronic case (with only a 1% stability).
With respect to the key expectations of the dissertation, it has been shown that print and electronic media do differ in the ways that they enable scientific communication, and this comparison was made possible by the juxtaposition of different metric analyses. The theoretical triad comprising Architectural – Network – Systemic dimensions has proven instrumental in maintaining coherence among the various analyses performed in this and the previous chapter. The differences in respective architectures, networks of keyword use, and systemic behaviour of the electronic communications revealed the print communications to be more codified than the electronic.
This chapter concludes the analyses of the internal dynamics of the SOEIS research group. We now turn to the externally oriented print and electronic communications. In the following chapter, the journal publication of the SOEIS group is examined as a primarily Mode I, print oriented communicative domain. That will be followed by a chapter dedicated to the comparison of the EuroCon-Knowflow mailing list with a
Introduction This chapter describes the analysis of the publication environment of the SelfOrganization of the European Information Society (SOEIS) research project. The publication dimension refers to the citation environment of the research project in 1996 (the year prior to the project’s inception), 1997, 1998, and 1999 (the years of the project), and 2000 (the year immediately following the project’s completion). Here, publications related to the SOEIS project are examined on both local and field levels and are analyzed as Mode I processes of knowledge production. Publications are characterized by formal constraints upon their communication, from the grammar to the typeset, and indicate a strong codification of discourse. These measures insure quality control. Indeed, nowhere is there a more salient example of this codified mode of communication than scientific journal literature. The SOEIS publication environment is assessed here to explain the secondary role of SOEIS participants, as academics – to publish their findings; this is done in order to expand our understanding of the role that specifically published print media has played in this process of knowledge production.
The journal publication analysis is best perceived in juxtaposition to the other analyses performed in this dissertation. The reader will recall that print communication (analyzed in Chapter IV) was linked with Mode I processes of knowledge production. By contrast, electronic communication (analyzed in Chapter