«Andrés Gregor Zelman The University of Amsterdam 2002 ii Mediated Communication and the Evolving Science System: Mapping the Network Architecture of ...»
In Chapter VI: Analysis of Journal Publication the question was posed: “what are the architectural parameters of the SOEIS publication environment with respect to its Cited and Citing dimensions, and to what degree does their overlap reveal a cognitive bias?” The subsequent analysis of the publication architecture showed that the citation environment of the SOEIS was predominantly oriented towards Policy and Informatics journals; this cognitive bias was evidenced by the centrality of Scientometrics and Research Policy. The publication architecture was further enriched by highlighting the European bias of SOEIS publications, and the consistency of coauthorship frequency which proved to be proportionately higher in the SOEIS group than in the core membership. In this way the SOEIS publications were shown to exhibit a particular architecture that operated as an information network in the Medium Theory sense of the term.
Similarly, the mailing lists analysed in Chapter VII: Analysis of Mailing List Environment were shown to operate as information networks, but there the emphasis was less on the cognitive orientation of the lists, as this was ‘given’ due to the restriction of the analysis to Science & Technology and Self-Organization Theory oriented mailing lists. Here it was asked: “can qualities common to all lists be identified, or do different lists perform specific functional roles in academic communication?” Prior to performing the analyses it was expected that the different lists under examination should vary with respect to the status of each as either project, intermediary, or field level lists. When compared, the lists were found to have discernable network architectures that were actively created and maintained; a direct correlation was found between list participation and thread participation, and the field level lists were found to perform roles particular to their expected function as evidenced by the participation levels of project, intermediary, and field level lists.
Given that the communicative domains differed so much in their composition, the examination of each necessarily entailed a different set of analyses. It was possible to compare the results of the analyses of the internal print and electronic communications, since their databases were so similar, but it remains a matter of comparing apples and oranges when the different architectural parameters of the external communicative domains of the SOEIS. Nevertheless, each was shown to operate as an information network in its own right, and they can be conceptually linked in this way. That is to say, the SOEIS research project certainly operated as a ‘social situation’ defined by a myriad of different patterns of information flow which collectively comprised the information network of the SOEIS. The analyses of the different architectural parameters of each communicative domain as an essential part of the collectively created and maintained information network of the SOEIS permitted an initial means of collating the results of these and subsequent analyses.
Where the architecture analyses used the metaphor of the information network to discern the overall parameters of the four communicative domains of the SOEIS, the network oriented analyses were generally more concerned with the dynamics of information exchange via these information networks. Chapter II: Theoretical Grounding introduced Actor Network Theory (ANT) as a discourse often used in the field of Science & Technology Studies to contextualize textual analyses. The ANT provided this study with a grounding for the analyses of the network properties of the respective SOEIS communicative domains under analysis. Whether comparisons between print and electronic keyword networks, citing and cited dimensions of SOEIS related publications, or threaded email communications, the ANT provided a soluble means of simultaneously appreciating the significant differences between the communicative domains while rendering a general network metaphor to bind the analyses. To assist with the interpretation of the ‘meaningful’ content of the communications analyzed, several meaning, or symbolic, oriented discourses from Structuralism, Poststructuralism and Structuration Theory were also reviewed. The dominant network metaphors obtained therein provided a number of different conceptualizations of where meaning could be ‘located’.
It was argued that where the structuralist discourse maintained a transcendent and relatively closed meaning system devoid of human context, poststructuralist approaches were shown to localize meaning in the social: in the act of communicating. Structuration Theory was then introduced to incorporate both approaches. Meaning is something that is both social in the sense of being created, but is also structural by definition – our operationalization of language through speech or writing systemizes the language structure. Spoken language leaves only traces in the mind, whereas the written word leaves traces that can be observed both synchronically and diachronically. Each discourse offered a network metaphor that proved instrumental to understanding the overlap of the four communicative domains of the SOEIS. The datasets under analysis were all necessarily ‘closed’, by definition;
the time periods selected for analysis were finite, but one can appreciate the evolutionary nature of communications geared to a collective end. Given the SOEIS as a research project with an advent and a termination, the datasets are best perceived as ‘closed’ in the structuralist sense of the term, but ‘open’ in the sense that the information content is expected to be different in each of the respective datasets and time periods under analysis. Hence, the SOEIS communicative domains were each assessed as time series.
For the analysis of the print and electronic communications the two years of the SOEIS research project were each divided into four periods of six months; for the analysis of journal publication the scope was stretched to five years to incorporate those publications which occurred before and after the SOEIS; and finally, the analysis of mailing lists measured the duration of threads over a finite period of analysis. Simply stated, the value of the symbolic approaches was to provide a sense of ‘meaning making’ in each of these communicative domains, thereby illustrating how each operated as an individual domain of knowledge production.
134 The first network analysis compared the fluctuations of keyword use over the four time periods of the SOEIS print communications, expecting that a discernable pattern of keyword reoccurrence would be visible. By comparing the top 50 keywords general topic words were found to be predominant and little was learned. Comparing each time period with the overall dataset proved more useful as particular emphases were found for each time period. The most impressive results of the keyword network analyses were found by comparing each period with the subsequent period, to reveal the transmission of word use over the four periods of the project as a process. Here it was found that words which concern the general functioning of the SOEIS research project appeared to increase in frequency over the time periods.
When the electronic communications were assessed as a time series, it was asked:
“does this distribution differ significantly from the results of the print analysis?” It was expected that the top electronic keywords would contain similar keywords as the print, but that a different emphasis will be located. Indeed this proved to be the case.
With the comparison of the top 50 keywords similar topics as the print dataset were predominant – hardly surprising given their shared participation to a common end.
Yet when the individual time periods were compared with electronic dataset, and with each other, it was found that keywords which appeared to supplement the activities of the SOEIS actually increased over time, whereas words which supplement the functioning of the SOEIS decrease. This is in stark contrast to the results of the print analysis where the occurrence of function oriented keywords appeared to increase.
Thus, internal communicative domains of the SOEIS were shown to be different; the difference between the print and electronic datasets can be partially explained by the role of each medium through the course of the project. The email communications were more process oriented, in terms of activity; rightly so, as it was largely by using this medium that SOEIS participants mutually decided upon meeting locations, introduced new ideas and suggested alternatives. Writing, as a formalized mode of communication, was shown here to be geared towards the functioning of the project and this can be understood with respect to the constraints placed upon communication by the pressure to create milestones and final reports.
The external communicative domains of the SOEIS were then assessed. Journal referencing activity and mailing list activity were subjected to different kinds of analyses, but were assessed with the same conceptual grounding. That is to say, both domains were unique in their manifestation, but were conceptually similar: both journal articles and threaded messages are written as responses to the literature relative to their respective domains, and are characterized by continued discourse. As representative of the internal dynamics of the SOEIS, print was selected as the exemplar of a Mode I oriented process of knowledge production and the electronic as more Mode II oriented. Similarly, journal publication is traditionally Mode I oriented, whereas mailing lists are Mode II. One must take caution, however, not to assume these to be mutually exclusive, since the dynamics of journal publication are certainly affected by the rise of new disciplines and hence new journals, as evidenced by the appearance of the Innovation literature in the citation neighbourhoods of Research Policy and Scientometrics, as shown in the publication systems analysis. Similarly, mailing lists were shown to exhibit traditional print oriented Mode I characteristics.
135 In Chapter VI: Analysis of Journal Publication, it was asked: “can networks of interrelationships be discerned by comparing the sum of most cited and most citing referenced journals? The SOEIS reference environment was shown to be policy and informatics oriented with the predominance of references going to Research Policy and Scientometrics. The SOEIS group was found to heavily cite Research Policy, while the group was itself cited primarily by Scientometrics articles which revealed a cognitive bias of the SOEIS group. It was shown that the publication environment exhibited a unidirectional flow of citation behaviour – Scientometrics was shown to heavily cite Research Policy. Research Policy is understood here to be the context of the application of analyses performed in Scientometrics; Research Policy is where scientometric studies are used and authenticated.
The network analysis performed in Chapter VII: Analysis of Mailing List Environment answered the question of whether Internet mailing lists differ significantly with respect to their list participation and threaded-ness. Differences were found between project, intermediary and field level lists and the examination revealed that the national lists did not operate in terms of threads. It was found that email communication via Internet mailing lists do foster unique network relations but also reinforce network relations associated with print media; a mixture of transition was found.
The SOEIS communications have been observed as unique domains through which knowledge has been produced. Self-Organization Theory was used, in part, to describe how print and electronic media use should be understood as mutually implicated phenomena as they have formed the single operation of the system of SOEIS communication. The systemic analyses performed on the four communicative domains analyzed treat each as a unique operation (of the collective SOEIS system), responsible in part for the recursive production of knowledge. By observing patterns in the observable phenomena, the analyses sought to identify the (largely unobservable) system of communication.
It was expected that an analysis of the print communications would reveal points of critical revision, showing path dependencies in the dataset. Each stage was expected to be instrumental to the furthering of the project communications in this medium.
The measure juxtaposed each time period with each successive time period for its expected information content. Contrary to the expectation, critical transitions were not found to be necessary for the communication to develop over the dataset. While path dependencies were not found in this analysis, it was shown that the transmission of words was better over the entire document set than the transmission over the document set of shared words. Similarly, for the systemic analysis of the SOEIS electronic dataset, no path dependencies were found to indicate critical revisions in the information exchanged. When the electronic dataset was subjected to an analysis of word transmission and specificity, and the results compared with the results of the print analysis, transmission was found to be better across electronic datasets, which was caused by the occurrence of words in some time periods and not in others.
Specificity remained lower in the electronic dataset. The systems analyses of the internal communicative domains of the SOEIS revealed that the print dataset achieved 136 the same communication as the electronic with less words, yet it was also shown here that the transmission of words was better in the electronic dataset than in the print.
Differently oriented analyses were employed to assess the systemic dimensions of the external communicative domains of the SOEIS. The systems analysis of the SOEIS publications analyzed the co-citation relationships between Scientometrics and Research Policy given their centrality to the citation environment of the SOEIS.
Using the relations between these two journals as the baseline as measured through
the factorial structure of their aggregated mutual citations, the question was asked: