«Andrés Gregor Zelman The University of Amsterdam 2002 ii Mediated Communication and the Evolving Science System: Mapping the Network Architecture of ...»
The electronic communications should appear more Mode II oriented than the print which exhibited Mode I characteristics. That is to say, the electronic communications should appear less codified by the parameters of the EU research project than the print communications. This difference should be observable through a comparison of the results of the print and electronic analyses. Importantly, these differences are expected to highlight ways in which print and electronic media codify information in distinctly different ways.
This question is informed by the Medium Theory notion of the information network, and aims to characterize electronic writing as contributing to a similar academic (information) network, but one that is expectedly different. With respect to research question one, the general expectation is that the electronic architecture will prove to be distinctly different from the print, and can be understood in this context as isolating the crux point between Mode I and Mode II types of knowledge production.
The second research question: are network properties of SOEIS electronic communication discernable by comparing the fluctuation of keyword use over the four time periods in the analysis, and does this distribution differ significantly from the results of the print analysis? As with the print analysis, insight into the cognitive orientation of the electronic communication is gained through examining keyword distribution in a number of different ways. Specifically: by looking to the top 50 keywords, by comparing the wordlists of each time period with the collective document set, and by comparing every time period with each other. The question is informed by Actor Network Theory and outlines the network features of electronic communication as individual communications that compile through collective interaction.
A key focus here is how individuals electronically contribute to produce texts and thereby form an information network. The primary expectation related to research question two is that the top electronic keywords will contain similar words to the print, but that a different emphasis will be located; specifically, it is expected that the electronic keywords will be more process oriented, rather than result oriented, given the use of the email communications to share and comment upon developments during the course of the research. It is therefore also expected that the electronic keywords will exhibit more of a speech bias than the print keywords, given the informal nature of email communications. The pattern of keyword distribution should prove to share certain qualities with the print dataset, and the notable differences will be interpreted in terms of electronic media bias.
Again, fluctuating changes in the position of the keywords are integral to understanding knowledge production as a process. The network dimensions isolated through ANT approaches and the theories of meaning from Structuralism, Poststructuralism and Structuration Theory provide a useful theoretical stance from which to understand the dynamics of meaningful exchange, and by extension, modes of knowledge production. Importantly, differences in the types of keywords used in the electronic database, by contrast to the print, should prove significant in the context of understanding the distinctions between formal and informal communications, by observing the variations in keyword emphases.
77 Fluctuations across the time series may be interpreted as evidence of changing cognitive biases as exhibited by their respective differences. In this way we can appreciate the means through which individual communications aggregate and reflect collective orientation. This reinforces the Actor Network Theory contention that textual analyses can indeed be conceptualized using network approaches, and positions the analysis toward the systemic aspects of the aggregate data sets.
Third, can we identify path dependencies in the SOEIS electronic dataset indicating necessary transitions in the information exchanged, and does this differ significantly from the results for the print analysis? This final analysis is contextualized using SelfOrganization Theory. Again, the systems oriented analysis compares the linear and non-linear associations between the time periods of the SOEIS Communications; in this case, from the electronic dataset. By comparing the expected information content of each time period as compared to the previous state of the communication, the analysis aims to determine if the communicated information followed particular pathways over others thereby indicating processes of critical transition. However, with respect to this last research question the expectation is that since no critical transitions were located in the print database it is unlikely that the electronic will exhibit any either. In the print analysis in the previous chapter it was discovered that there was no apparent self-organized criticality in the print dataset; when compared with the results of the system oriented analysis of the electronic dataset additional perspectives may be attained. The specificity and transmission of the words in the electronic dataset is examined to expand our understanding of the differences between the print and electronic data sets.
In Chapter VIII: Integration & Conclusions, the results of this comparison between print and electronic modes of knowledge production will be described with the results of the analysis of SOEIS journal publication and the analyses of the EuroConKnowflow and Self Organization of the Information Society (SOIS) mailing list thread behaviour (covered in Chapters VI and VII, respectively). In what follows, the results of the electronic and print textual analyses are compared in tandem.
Architecture The analysis of the electronic data set entailed an initial filtering of the email data to eliminate redundancy.1 A total of 1261 emails were compiled into four 6 month chunks denoting the same four time periods as the previous print analysis the texts were run through the WordSmith program to obtain a range of descriptive statistics including the number of emails, relative size, word count, unique-word count, the percentage of unique words and their mean ratio percent. The texts were then filtered 1 The texts were filtered using three logics. First, the redundant lines of the respective emails were eliminated (next, previous, and re: subject lines), as were all signature files. Second, all messages which were a response to another message had the previous message removed. Finally the texts were filtered using an adapted stop-list (see Appendix B.1).
78 using the same adapted stop-list.2 Table 5.1: Electronic Architecture shows the basic statistical information.
In the electronic database it is observed that email activity decreases over time, suggesting an externalization of the larger SOEIS group.3 By contrast, document size, word occurrence, and unique word occurrence all increase over time with the notable dip in the third time period. This dip may be due to the combination between the two lists (EuroCon-Knowflow and SOIS), as the third phase of the electronic communications would have been a transition period for the participants of the SOEIS research project. The mean ratio percentage of unique words shares this dip in distribution, whereas the percentage of unique words decidedly increases over time.
As with the print analysis, the percentage of unique words is interpreted here as an indicator of the style of variation between unique words and total word occurrence, as distinct from the mean ratio percentage of unique words which is recomputed every 1,000 words.
The meaningful difference between the print and electronic architectures is discernable through a comparison of the mean ratio percentage of unique words over the four time periods. Where the mean ratio percent remained relatively constant over the four print datasets, there was a marked increase in mean ratio percentage over the four electronic datasets. This difference may suggest an accretive influx of new words and therefore new types of research interests. By implication this would mean that the print document set exhibits a more codified communication than the electronic; new ideas appear to be communicated electronically, not in the print communications.
The results from the print analysis revealed evidence of codification as both a process, perhaps as the result of a single author or group of authors acting as editors prior to the project submission, and an a priori codification. The increasing mean ratio percent in the electronic dataset suggests precisely the opposite – there appears to be evidence of resistance to codification and an embracing of new research questions, thereby confirming the expectation that an analysis of the architectural features of the electronic dataset would exhibit a different codification style than the print.
2 The four 6 month chunks delineated in the print analysis to correspond with the four major developments of the print communications dataset: Application, Milestones, Reports and Final Results. The electronic communications were divided into the same four six month chunks to provide a basis for comparison.
3 The decrease in email activity is interpreted as an externalization of the SOEIS group; the main thrust of project communications occurred in the print dataset. The introduction of the SOIS mailing list was a strategic move meant to expand the communications of the EuroCon-Knowflow mailing list to include non-European elements of the information society, but in effect reduced the flow of electronic communication.
79 Network The network analysis of the electronic dataset involved an examination of the relationships between keywords over the four time periods represented by the four wordlists (E1, E2, E3, and E4). Network properties are exhibited by identifying fluctuations in keyword and collocate distribution over the four time periods. Shifts in keyword emphases over the course of the project are expected to provide evidence of changes in the collective cognitive organization of the project.
As shown in Chapter IV: Analysis of Print Communication, the keyword network analysis performed on the four wordlists involved a range of distinct approaches. The first isolated the top 50 occurring keywords for each time period; the second identified positive and negative keywords by comparing the four respective texts with the electronic dataset; the third compares each text with each other (not the full electronic data set), and finally, keyword collocates are located for a selection of representative keywords.4 In this way major developments in the concepts being exchanged are outlined.
Ten keywords from the top 50 most commonly occurring words in each time period in the electronic data set were selected on the basis of their relevance to the SOEIS research project. They are displayed below in Table 5.2: Top Electronic Keywords.
The keywords and their distribution demonstrate the general research environment or context of the SOEIS group as similar to that generated through the print communication. Here we find a considerable overlap between those words found in the print and electronic databases; half of the top ten words shown here also occurred with high frequency in the print dataset: Information, Networks, Organization, Project, and Task. Interestingly, they generally appear to occur with similar frequencies across the time periods irrespective of the medium in which they were generated (print or electronic), with the exception of Networks and Organization which occur less frequently in the last period of the electronic dataset than in the print.
Further, the keywords Information, Networks, Organization, Project and Task each occurred with high frequencies in the last time period of the print data set, whereas in the electronic dataset they occur with significantly lower frequencies, with the notable exception of the keyword Information.
4 Only those words that occurred within five words to the left or right of the query word were included in the correlate analysis; too narrow a window limits results and too wide a window dilutes results.
80 Five keywords are unique to the top 50 of the electronic dataset: Research, Conference, Knowledge, Meeting and Policy. This selection of words reveals a slightly different emphasis of the research project in its use of the email lists. With the exception of Knowledge, it is arguable that all of these words imply some sort of activity that arguably supplements the research project itself. Also notable is that Meeting, Task, and Networks occur relatively infrequently in the last time period of the electronic dataset. Indeed, it is logical that the words Meeting and Task would have been used less in this medium when the project reached completion. This confirms the expectation that the keyword distributions in each medium are related to its designated function.
Because the most commonly occurring words generate noise and cloud the evidence of critical keywords, a second analysis was performed whereby critical keywords were isolated by comparing texts. Two distinct keyword analyses were performed on the electronic dataset whereby the respective wordlists (E1, E2, E3, and E4) were compared with the full document set (E-All), and then compared with each other (E1 with E2, E2 with E3, E3 with E4). In this way keywords for each time period were determined on the basis of how frequently the keywords were used over the whole project by comparison with the full document set, and then again by highlighting the transition between the time periods.
Again, the shared keywords revealed by comparing wordlists using the WordSmith program are assigned either positively (+) and negatively (–) thereby representing those words which occur more or less often than expected, given the texts that are compared. The equal (=) signs simply represent that the words were not designated as key, not that they did not occur. The comparison of the individual wordlists with the full document set isolated four lists of keywords, one for each time period. Fifteen salient examples are shown here; these were selected on the basis of their relevance to the research project, and by extension, the perceived supplementary role of the electronic medium. Table 5.3: Electronic x All; for the results of this analysis.