No

This paper intends to promote an awareness of the overall research output in library and information science in South Africa, and is based on a publication count and analysis of peer refereed articles indexed in the LISA and Thompson Scientific (formely ISI) Web of Science databases (SCI,SSCI,A&HCI) between 1993 and 2006, using journal, subject and author indicators for the analysis. The recommendations are in favour of expanding the publication threshold by diversifying the output such that it includes currently marginalised domains.


I. Introduction
Whereas there is no consensus on the best way to measure research output in a given discipline, most members of the scientific community, particularly those in favour of quantitative measures of research (see ISSI conferences), concur that peer refereed journals offer a verifiable platform/source of measuring the research productivity of scholars.Even in this area, there is a strongly held view that the journal impact factor [of peer refereed journals] (e.g.determining the degree of cited-ness of articles in a journal) should be used to determine the most important and influential research journals and research papers/articles in a discipline.The Citation Impact Factor (ClF), proposed by Eugene Garfield in 1969 (Garfield, 1994 :41 I), is defined as the average number of citations in a given year of articles published in a journal in the preceding two years.Normally, citations received in one year are divided by papers published in the two previous years in order to obtain the ratio.The approach used to determine the quality of research has therefore not been uniform.Evidently, there are those who are in favour of qualitative measures of research (e.g.Gorman 2000, Calvert andGorman 2002) and also strong proponents of peer review as a measure of research quality (e.g.Harnad 1995 3 ).Similarly, there are those who are in favour of citation analysis and the journal impact factor as a.quantitative measure of research output (e.g.Garfield 1971Garfield , 1972Garfield ,1994Garfield , 1998)).For example, when defending qualitative measures of journal quality as opposed to quantitative measures based on cited ness or the impact factor, Calvert and Gorman argue that "The fact that paper x is cited y times is not an indicator of quality, but rather that it is cited -it is available, it is in the journal held by many libraries, the author (or publisher or editor) is particularly good at self-promotion" (Calvert and Gorman 2002: I).Harnad has always provided peer review with overwhelming support and defence.In one of his seminal articles on peer review he (Harnad 1998: paragraph one) argues that journals should not be free from the "process of peer review, whose 'invisible hand' is what maintains its quality".Although other forms of research output, such as books, conference proceedings, reviews, theses and dissertations, patents, and other research reports of limited circulation are used to measure research output, journal articles are still the most dominant, favoured and easily verifiable for quality control in scientific research.Each country, and in some cases institution, determines its research quality in different ways.For example, a quality research output in South Africa will appear in a prescribed list of 255 South African Journals 4 , Thompson Scientific (151)databases s and IBSS databases 6 , and will not include correspondence with the editors, abstracts or extended abstracts, obituaries, book reviews, news articles and advertorials.For each article published in such a journal, a substantial government research subsidy -which in itself is regularly revised and increased -is paid to the author's affiliate institution, which then decides on how to share the subsidy with the authors/contributor.
The first part of this analysis was based on the output of graduate (masters and doctoral) dissertations and theses from 1993 to 2000, as reported at the 66 th IFLA conference held in Jerusalem (Ocholla, 2000).The variables included gender, language, population group, institutional affiliation, subject, and the quantity and output of both masters and doctoral theses over that period.It was observed that the preponderance of theses was produced at masters level in the English language by women, and that the universities of Natal [now KwaZulu-Natal] -Pietermaritzburg campus, Pretoria, and the Rand Afrikaans University (now the University of Johannesburg) lead in productivity.Additionally, the multidisciplinary nature of information science exhibited elements of boundary crossing, collaboration and borrowing from computer science, business management, geography, music and political science in graduate research output.Although this analysis has not been extended to 2006 due to the closure (in 200 I) of the unit previously indexing research output at Potchestroom University (now part of the University of the North West), the productivity pattern reported by Ocholla (2000) has not changed much.However, there are marginal variations, for example other universities that did not feature well in that study (such as the University of Zululand) have made significant progress during the last six years, more publications are emerging from the formerly marginalised communities largely through co-publication with established researchers/postgraduate masters and doctoral research supervisors.

Bibliometric/lnformetric
studies are widely used to inform policies and decisions in political, economic, social and technological domains affecting information flow and the use pattern within, between and outside institutions and countries.Although Library and Information Science (LIS) studies of this nature solve problems related to collection development, information retrieval, systems design, user studies, management, and knowledge organisation, among others, in Africa bibliometric studies are limited.Those focusing on LIS are insignificant, with the exception of a few studies reported largely by West African scholars such as Aina (1998), Aina and Mabawonku (1997), Aina and Mooko (1999), Alemna andBadu (1994), Alemna (1996;200 I), Kadiri (200 I),and Mabawonku (200 I).There are a few noted studies in South Africa by Boon and Van Zyl (1990), Ocholla (2000: 200 I) and Ngulube (2005a;2005b).This study adds to the cited studies by providing, in general, an awareness of the overall research output from within the Library and Information Science discipline in South Africa based on a publication count of peer refereed articles appearing in national and international LIS journals, specifically those indexed in LISA and lSI databases.This is in order to determine whether diversification and output with regard to authors, journals and subject coverage and research collaboration has occurred over the period.The paper therefore attempts to address the following questions: In which journals do the LIS authors  1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 Total

Author publication pattern
The analysis includes non-South African residents who have either produced co-authored articles with South African residents, published in South African LIS journals, completed their masters or doctoral studies in South African academic ii1stitutions, or lived/resided in South Africa before but migrated to other countries, as captured for the master list.As presented in Table 2,    research topics for research partnership and graduate enrolment (see Ocholla 2000).It is recognised that there is no universally acceptable classification scheme of LIS by subject, and the bold attempts by Jarvelin and Yakkari in the last decade (see Rochester and Yakkari 1998) to establish a classification scheme or taxonomy have not been without criticism.Attempts to obtain usable subject taxonomy from LISA and lSI were unsuccessful,as LISA does not seem to offer one, while lSI categorisation (e.g.library science, information science, computer science or information systems, etc) is too broad.A subject descriptor in Library and Information Science Abstracts was therefore used to select the main/broad subject area arbitrarily for the analysis,as reflected in Table 3. been used in the analysis.Researchcollaboration has a number of benefits, as outlined by Katz & Martin (1997).Among them, according to the authors, are: that colliboration enables researchers to share skills and techniques, and is one way of transferring knowledge (especiallytacit kn6wledge); through clashingviews it may bring about the cross-fertilization of I ideas, which may in turn generate new insights or perspectives that individuals, working on their own, would not have grasped; collaboration provides intellectual cbmpanionship (i.e.within a practising community); collaboration plugs the I researcher into a wider contact network in the scientific community; and it enhancesthe potential visibility of the work.Thus, collaboration helps speed up problem s~lving, stimulates creativity and enables inter-disciplinary boundary crossing, which in turn enriches knowledge developm~nt and transfer.
A total number of 145 South African auth6red articles were published either by single authors or co-authored.Of the I 145, individual/single authored were 45 (31 %), two authors appeared 78 (53.8%) times, three authors 17 (I 1.8%), and four authors 4 (2.8.%) times.There was on~instance in which a single article (0.6%) was published by 20 authors -an internal co-publication from the University 6f Pretoria.As to whether collaborative publication was internal, external, external but within South Africa, or external tiut with foreign countries, it turned out that of the 100 co-authored articles, I 55 (55%) were internal (i.e.published by colleagues from the same institution), and 45 (45%) were external (published with colleaguesfrom other institutions).Exte~nalco-authorship with South African Institutions came to 23 of 45 (51.2%), external but with non-South African institutions totalled 20 of 45 (44.4%), while external but involving both South African and foreign institutions produced 2 of 45 1(4.4%).Figure I and Table 4 shows the nature and type of research collaboration through single or multiple publi~ations.Evidently, there are more co-authored articles (69 %) than singleauthored articles (31%).Furthermore, therJ is limited external (45 %) collaboration within and outside the country.
Even collaboration between institutions within the country is just slightly more than half (55%) of all collaborations.

Conclusions
South African LIS researchers/authors largely publish in local journals (46,3%), led by South African journal of Library and Information Science-SAjLlS (25.1 %), Mousaion (I 1.9%), and the South African journal of Information Management (9.3%).Publication in Thompson Scienctific/ISI and International Bibliography of Social Sciences (IBSS) indexed journals is also on the increase, based on similar recognitioh and rewards.Notably, while more South African based researchers publish I in peer refereed LISA indexed LIS journals (250 authors), publication in lSI indexed journals (67 authors) is limited.The average citations (2.2 ratio) originating from\the 256 articles is insignificant, although it differs less from the citation ratio in LIS alone and when compared to other social science disciplines, as observed by Onyancha (2007).Onyancha also reveals that although South Africa publishes most 6f Africa's LIS research, it receives comparatively fewer cites for its articles.In terms of subject orientation, there Iis an impressive diversification and research focus on core areas of LIS education, such as management, information retrieval, services and dissemination, and the application of ICTs.Formidable niche areas seem to have been created by bstablished academics, who continue to encourage more young researchers and publications in their fields of specialis~tion.However, we do not believe that this trend could lead to an oversaturation of specialists in a particular field Jt the expense of other less attended research domains.We believe that in a nascent democracy such as South Africa, sp~cialisation should go hand in hand with diversification in order to enable the I creation of capacity in marginalised fields.There is therefore potential for accelerated and enlarged publication output in the discipline in South Africa, provided thJt: novice/potential researchers (such as postgraduates) receive publication support from research supervisors, the g6vernment continues to pay subsidies to institutions based on accredited publication output, and institutional perfo~mance measurement indicators emphasise publication output.Thus, both I quantity and quality can be maintained.The results relating to popular research topics have been compared to international trends reported by, for exampl~, Maxine Rochester and Pertti Vakkari (1998).
Research collaboration as observed thrdugh co-authorship (69 %) is encouraging, as the bulk of such collaboration increasingly occurs between the research s~pervisor (of largely masters and doctorates), and the postgraduate student, I who tends to be a member of the staff/faculty from the supervisor's academic institution.However, it was observed that I inter-institutional research collaboration within South Africa is average (5 1.2% of 45), and more or less similar between I South African and non-South African institutions (44.4%).We believe that inter-institutional research and international research collaboration can reap benefits froin the research collaboration currently going on within the Dissanet 8 project, which focuses on promoting LIS research c1ollaboration in South Africa.The increased research collaboration between I established researchers and novice researchers and postgraduate students is commendable.We conclude that, since South Africa still leads in research and pUblicktion output in Africa (see Onyancha 2007), the rapidly growing research and publication output and support in the count~offers promising opportunities for research and professional collaboration that could be explored and exploited beyon~South Africa's borders.This study is not inclusive or conclusive, lasit only focused on research publication output appearing in peer refereed I journals indexed in LISA and lSI Web of Science (SCI and SSCI) between 1993-2006, for reasons discussed in the introduction (section I).Other parts of the! analysis, such as subject orientation and research collaboration (LISA is left out), are also incomplete.The study does not measure individual or institutional research output, which is more I complicated (i.e.requires more variables).The question stands as to whether publication output in peer refereed journals can be used to measure/determine researdh output in a discipline such as LIS.We believe that an inclusive research agenda covering research quality, quantity, c611aborationand diversification needs further exploration.

I
Perhaps most articles are published in South African LIS journals (and other journals) none of which is indexed by lSI, or due to other reasons earlier cited from CalJert andGorman (2002: I).

Table 3 Subjectorientation
The last part of this study focused on collaborative research output.Only data from the 256 records in lSI have thus far I

Table 4
Type of Collaboration