Russian Journal References and Citations in the
Science Citation Index Databank.
Eugene Garfield, Ph.D.
Institute for Scientific Information
Philadelphia, Pa. 19106
Presented at a U.S./USSR Symposium on forecasting information requirements
Yale University, New Haven, CT.
October 20-23, 1974
This paper reports an analysis of citations of and by Russian journals. The citations analyzed were extracted from the Science Citation Index (SCI) data bank.
The term Russian journals is used here as it is daily used in. libraries in the United States. We are aware of its inadequacy and inaccuracy, but plead its convenience. A few of the journals are Slavic, but not Russian. The term Soviet journals might seem more appropriate, but it would not be. An important group of the journals considered is published outside the Soviet Union--the so-called translation journals. Neither Russian nor Soviet, they are nevertheless the product of Russian and Soviet research. They also present, as we learned in this study, a formidable stumbling block in journal citation analysis of this type. I speak here only in terms of statistical bibliography as regards the translation journals. The larger question of their use and effectiveness is indeed a complicated problem. At the moment I believe one cannot give much more than a subjective evaluation of whether, or how well, the cover-to-cover translation of journals serves its intended purpose of providing a ‘permeable membrane’ between the Soviet and other linguistically separated scientific communities. Though they undoubtedly make for confusion in citation analysis, citation analysis should be able to contribute some partial but objective measure of their effectiveness.
Materials and Methodology
The source of the data analyzed for this report is the magnetic-tape data bank compiled since 1961 for compilation of the annual volumes of the Science Citation Index. By the end of 1972, the data bank contained more than 36 million citations.
All these citations appeared in the references of articles published in journals processed for the SCI. Journal references, of course, cite a wide variety of published items--monographs, patents, theses, abstracts, proceedings of meetings, symposia, and conferences, published and ‘unpublished’ reports of official and private organizations, and other journal articles. Journal articles account for most of the citations made in references in journals. Over the years 1961-1974, they have steadily averaged almost 80% of total citations in journals processed for the SCI.
In 1970, the Institute for Scientific Information undertook a citation study of the data bank as it stood at the end of 1969. Preliminary results of that initial study have been reported in Science1. There we at tempted to show what journals were most frequently cited in 1969 by the scientific journal corpus as a whole, and what journals had published articles with the most significant impact, that is, what journals had been most frequently cited in 1969 in relation to the number of articles published in 1967 and 1968.
The methodology of that initial study has been used here, and for numerous studies of groups of journals in different specialties.2 It may be useful to recapitulate briefly the steps involved.
In our initial study we extracted from the data hank all citations from references in journals processed for the SCI during the last quarter of 1969. This reference/citation file was then arranged to show what journals each journal had cited, and how frequently, and by what journals each journal had been cited, and how frequently. In each case, a journal’s citation count (as source or citing item and as reference or cited item) was distributed by publication year of the items cited.
Various listings of this manipulated file make up ISI’s Journal Citation Reports (JCR)3.
As planned in 1969, we have extracted from the data bank all 1972 citations, and arranged and listed them in the manner described to show citing and cited frequencies. This 1972 file will be the first of a series of annual editions of 3CR.
For this study we extracted from the two files so far created all citations from references in Russian journals, and all citations of Russian journals, whatever their or gin. In the manner described, these two groups of citations--those from the file covering the last quarter of 1969, and those from the file for the entire year 1972--were separately arranged to show for each year what journals cited Russian, journals most frequently, and what journals were most frequently cited by Russian journals.
In other words, we have treated Russian journal references and citations of Russian journals from the SCI data bank in terms of one huge imaginary Zhurnal Sovetskikh Nauk.4
The results of this extraction and manipulation of citations are given in Figures 1-4. Figures 1 and 2 show cit at ion frequencies for 1969. Figures 3 and 4 show citation frequencies for 1973. Figures 1 and 3 are analogous. They list the seventy-five journals that most frequently cited Russian SCI source journals in 1969 and 1972 respectively. Figures 2 and 4 are similarly analogous. They list the seventy-five journals that were most frequently cited by Russian SCI source journals in 1969 and 1972 respectively.
As noted above, the first of the extracted files included only references/citations from journals processed during the last quarter of the year 1969, whereas the 1972 file included all 1972 citations. As previously reported1, the 1969 quarter-year sample was tested for statistical validity by comparing it with a file created by extracting every twenty-seventh 1969 reference/citation. With that test as background, we have sometimes in studies of this type extrapolated the quarter-year data to facilitate comparison with the 1972 data. In this case we have not clone so because of the special problem presented by overlapping of original and translated journals: I believe comparison will be to an adequate extent facilitated by rank of journals on the lists and by the relationships expressed in the figures in terms of percentages
Limitations and Cautions
The number of source journals in the SCI has grown steadily over the years since 1961. In 1961, there were only 613 source journal s by 1966 the number had increased to 1573; by 1969, to 2180; by 1972, to 2425. The number of Russian source journals has increased, and at a rate relatively greater than the increase overall. In 1961, 1.6% (10) SCI source journals were Russian, all original versions. In 1966 2.2% (35) were Russian; all of them except one (the Doklady) were translated versions. In 1969, 2.4% (53) of the source journals were Russian; all except six were translations. In 1972, 3.4% (83) were Russian; about half originals, a quarter translations, and a quarter in both versions.
The size of journals varies considerably, and number of journals does not equate with number of articles, much less with number of references and citations, in 1961, Russian source journals contributed about 3% of the source articles; in about 4%; in 1969, about 5%; and in 1972, about 7%.
If only to stress our awareness of the fact, let it be noted that the 102 Russian source journals of the 1972 SC cannot be claimed to constitute the "Russian scientific and technological literature." We believe, however, that, these 63 journals include most of the significant Russian literature. What is true of the scientific literature as a whole is true of national and specialty segments of it, as our numerous journal citation studies have shown. In every case, a relatively small core journal accounts for most of the articles, most of the references, and most of the citations. The lists in Figures 1-4 support this belief in the case of Russian journals. All Russian journals that appear in these lists ––that is, journals that most heavily cite and are cited by the Russian literature and those that most heavily cite and are cited by SCI source journals as a whole–– are themselves SCI source journals, in either their original or translated versions.
In addition, it should be remembered that the lists of most-cited Russian journals (Figures 2 and 4) arc not restricted by the methodology to SCI source journals. Except for a journal whose citation is overwhelmingly self-citation, any journal highly cited by the study group would have turned up for inclusion in the list, whether or not a source journal. As a matter of fact, we rely upon this phenomenon to reveal lacunae in our coverage. When such journals do turn up, they arc the first to be considered for addition to the SCI. All the Russian journals in Figures 1-4 are source journals because the SCI does include most significant Russian journals in its coverage.
One problem deserves the closest attention. I have mentioned the impediment to citation analysis posed by the existence of the translation journals. That impediment is citation of original Russian versions by their translated versions. For example, most articles in translated journals will cite the original. Considering that the average article is cited only 1.7 times a year, the effect of that single pseudo-citation cannot be dismissed as trivial. More serious than this, however, is the use of both genuine and pseudo-citation by other articles in the same or other translation journals. For example, the translation journals of the American Institute of Physics give a translation of the references and of the citations each contains. They give also, however, citations of the translated articles that appeared in the Soviet Physics series.
It is not as simple as it would seem at first thought to determine the extent to which citation bf Russian journals, in counts like ours, has been inflated by the pseudo-self-citation and the pseudo-citation within the double citation. We do not know how often non-Russian journals follow the practice of the Soviet Physics series when citing Russian articles, nor how often authors may actually use a translation, but cite the original Russian.
Thus, unless considerable care is exercised in examining the citing and cited records of individual journals, the counts will be heavily inflated, to an extent difficult to determine, and perhaps impossible without examination of thousands of individual bibliographies.
In the lists in this report we have combined the citation records of an original Russian and its translated version, but we have discarded those pseudo-citations we have been able to detect with fair certainty.
As mentioned previously, we are confident that citation analysis can contribute to evaluation of the effectiveness of cover-to-cover translation. The detailed cited and citing lists for individual journals allow one to a certain extent to differentiate between use of an original and of its translation. That differentiation is beyond the scope of this report, and will be the subject of later research.
Though it is largely the case, one must recognize that an analysis of Russian journals only cannot account for the bibliographic impact of all of Soviet science and technology. Many highly cited articles by Soviet scientists have been published in foreign or international journals. I suspect that their impact has been greater than if they had been published in Russian journals. In a subsequent study, we intend to identify these articles. They can be accessed either by their authors’ names or by their authors’ addresses and affiliations.
Some Preliminary Observations
In studies of this type, considerable effort must he spent in unifying variant abbreviations of the same journal titles. In the course of this study, it became quickly apparent that translation journals do bibliography a major disservice in their treatment of the titles of Russian journals. Translating a title may be a necessary evil, but changing can surely have no excuse. Any American researcher would be dumbfounded, scanning an issue of Sovetskaya Meditsina, to come across the ‘title’, whether or not abbreviated, Amerikanskaya Meditsina JAMA. That same researcher, however, along with the rest of us accepts Soviet Physics JETP without a thought. Any source of bibliographic confusion is to be avoided. Soviet Physics JETP is not the title, nor even a translation of the title of Zhurnal Eksperimentainoi i Teoretika Fizika. I would strongly recommend that Soviet publishing houses include in their commercial agreements on translation a prohibition against such retitling of Russian journals. I should like to see all translation journals adopt the practice followed by Current Contents. The title of the Russian is given in Russian, then transliterated, then translated. As an alternative, there is Stal in English, which follows the excellent example of Angewandte Chemic International Edition.
Not all the bibliographic confusion we have encountered can be laid at the door of translation journals, however. Russian journals must accept blame for some it. Many tend to follow a practice I have previously noted in the case of English-language astronomy and astrophysical journals,5 the use of initials and acronyms as abbreviations It takes an experienced eye to recognize in roman or Cyrillic the significance of ZIIOKII, FTT, FMM, TVT, etc.
Observations on the Lists of Highly citing and Highly Cited Journals
The legends of the figures indicate briefly what the columns of figures report. A few comments on the reason for their inclusion may be useful. Each list gives the total number of citations made or received by a journal, the total number of citations made or received by the Russian journals alone, and, of the latter, the number of self-citations. The relation of these counts to one another is given in percentage form. The relationships are useful in interpreting the ranking of journals on the different lists. A final column in each figure gives the journal’s impact factor.
A journal’s impact factor is determined by dividing the number of times it was cited in a particular year by the number of articles published during the two preceding years. Thus, the impact factors in Figures 1 and 2 for 1969 represent the frequency with which the "average" article published by the journal in 1967 and 1968 was cited in 1969. Similarly, the impact factors in Figures 2 and 4 represent the frequency with which the average article published by the journal in 1970 and 1971 was cited in 1972. A discussion of various methods of calculating impact, and of our choice of this particular method, will be found in the Science article cited above.
Impact factors apply particularly to a journal’s citedness, and thus will be of more value in interpreting the lists in Figures 2 and 4 than those in Figures 1 and 3. They are useful however in the latter case as well.
In Figure 1, for example, the Doklady is shown to have produced more citations in 1969 than any other journal in the Russian group. Only 27.5% of its citations were citations of the Russian journals in the group, and of those 39% were self-citations. The impact factor is relatively low, 0.572. Compare the Doklady with the second journal on this list of journals that cited the Russian group most often, Fizika Tverdogo Tela. It produced 3704 citations in 1969, of which 50% cited the Russian journals, including a self-citing rate of 60.0%. The Doklady, as we know, is much less specialized than Fiz. Tverdogo Tela, and the figures confirm the fact. The, latter depends much more heavily on Russian journals than the Doklady, and its rate of self-citation indicates high specialization. The high self-citation might also indicate extreme parochialism. The high impact factor, 2.046, suggests that such may not be the case. The detailed listing for this journal is in ISI's Journal Citation. Reports 1972 show that in 1972 it was cited by at least 1450 other journals.
A main purpose of this study was some measure of "cross-citation" between the Russian and other scientific literatures ––to measure, in other words, how frequently the Russian literature cites outside itself, and how frequently other literatures cite it, and to compare the findings for 1969 and 1972. Even considering possible aberrations due to the extrapolation described in compiling the lists in Figures 1 and 2, basic consistencies between the records of the two years are remarkable.
In Figure 1, 32 of the 75 journals that cited Russian journals heaviest in 1969 are non-Russian. In Figure 3, only 17 of the 75 journals that cited Russian journals heaviest in 1972 are non-Russian. The drop from 32 to 17 may be due to the increase of Russian source journals between 1969 and 1972, from 53 to 83. But it is interesting to note that the number of non-Russian citations of the Russian journals decreased hardly at all. In 1969, non-Russian journals accounted for 17% of the citations made by the 75 heaviest citers of the Russian SCI journals. In 1972, the corresponding figure is 15%.
It is interesting to compare these figures for citations of the Russian source journals with information from Figures 2 and 4, which give citations made by the Russian source journals. In Figure 2, the list shows that 36 of the 75 journals most heavily cited by the Russian source journals were non-Russian. In Figure 4, the number of non-Russian journals is about the same, 35. In 1969, however, the 36 non-Russian journals accounted for 59% of the citations listed. In Figure 4, the 35 non-Russian journals accounted for 77%. The 1972 file shows a decided increase in non-Russian citations by Russian source journals. One is at least tempted to speculate about the cause of this increase. As noted above, the number of Russian source journals in the SCI increased between 1969 and 1972. The simplistic conclusion that this increase would have resulted in more Russian citations is proved false. It may be that we were not, in the case of the 1969 lists, as successful as we hoped in ‘deflating’ Russian citation counts. But in view of the attention given that problem, I cannot credit to it the increase of almost 20% in non-Russian citations made by Russian source journals. There is the further fact that the make-up of the two citing lists (Figures 1 and 3) and of the two cited lists (Figures 2 and 4) has changed relatively little, especially when citation percentages are considered. About 57% of the journals are common to Figures 1 and 3, and to Figures 2 and 4, and these journals are for the most part concentrated at the top of the lists.
In this connection, items 13 and 15 in Figure. 4—journals most cited by Russian source journals--are extremely interesting. Item 13 is Nature and item 15 is Journal of Biological Chemistry. Thirteenth in 1972, Nature moved up from position 28 in 1969. Journal of Biological Chemistry does not appear at all among the list of 75 journals most heavily cited by Russian source journals in 1969.
To me, the increased importance of Nature and the appearance of the Journal of Biological Chemistry are at least signals of a pattern that deserves, I believe, further consideration. All four lists are heavily oriented toward physics, chemistry, and their technologies. In Figure 3, and more markedly in Figure 4, biomedical journals, both Russian and ‘foreign’, are much more in evidence. I am fairly certain that this phenomenon cannot be attributed to either methodology or changes in SCI source journal lists. It seems to reflect accurately a shift in emphasis within Soviet research. It will be interesting to learn whether other ‘research indicators’ confirm this result of a wholly algorithmic and objective procedure.
I have previously mentioned that ISI’s Journal Citation Reports give detailed listings of citing cited relationships for every source and reference (cited) journal, and that these detailed listings show a distribution of cited -items by year of publication. We know from yearly statistical analysis of SCI input, that articles tend to be cited most frequently one to three years after the year of publication. Every year since the SCI began publication in 1961 about 30% of the citations processed were from one to three years old. The increase in the number of source journals has had no effect at all upon this constant, just as it has had little effect on the number of times the average article is cited each year, about 1.7. (I have succumbed to an understandable temptation and will soon introduce the latter to the documentation literature as Garfield’s Constant. My preoccupation with 1.7 has perhaps diverted deserved attention from the equally constant 30%).
I have introduced the one-to-three year 30% citation constant at this point to suggest its possible usefulness in evaluating the effectiveness of translation journals. If translation journals are vital to the flow of scientific and technical information from Soviet to other research, and if their users regularly cite translations instead of originals, one should be able to detect a variation from the 30% constant when considering the citation of Russian journals by English-language journals, and perhaps others as well. I base this assumption on the fact that most translation journals appear a not inconsiderable time after the originals. In the few cases I have examined, the considerable variation from the constant suggests a need for further study.
For example, detailed listings from ISI's JCR 197.2 show the following: in the case of Zh. Eksp. Teoret. Fiz. (Soy. Phys. JETP), the one-to-three year, after publication citation percentage is 38%; in the case of Zavod. Lab. (Industr. Lab.), 36% in the case of lisp. Fiz. Nauk (Soy. Phys. lisp.), 35% in the case of Auks. Zh. (Soy. Phys. Acoustics), 25% and in the case of Zh. Neorgan. Khim. (J. Inorgan. Chem.), 24%. Whether these variations are significant will require extensive statistical analysis. Such a study would certainly be worthwhile. We have assembled the data for the interested researcher.
In conclusion, I should like to draw attention to the impact factors for the listed journals. I have explained its calculation above, and previous studies of journals and journal groups have, I believe, established it as a reliable criterion in journal evaluation. Overall, there are few startling differences in impact factors between the journals common to the two highly cited lists in Figures 2 and 4. In some cases, such as Science (2.894 in. 1969; 4.399 in 1972), the change is due primarily to omission from ‘published items’ of correspondence, book reviews, etc., whose inclusion had worked to the disadvantage of, such general journals. I find it interesting that journals with a higher rank in 1972 than in 1969 in the heavily cited listings of Figures 2 and 4 usually have a higher impact factor ––if they are Russian journals. Oddly, that does not seem to hold true for ‘foreign’ journals cited by the group of Russian journals.
References:1. back to text Garfield E. Citation analysis as a tool in journal evaluation. Science 178:471-79, 1972. Reprinted in Current Contents No. 6, 7 February 1973, p. 7-24.
2. back to text See, for example: Garfield E. Journal citation studies. 21. Engineering journals. Current Contents No. 27, 7 July 1975, p. 5-10. -- A list, of these journal citation studies, and of related highly cited article studies, is available from the author on request.
3. back to text Garfield, E. The new ISI Journal Citation Reports should significantly affect the future course of scientific publication. Current Contents No. 33, 15 August 1973, p. 7-8.
4. back to text Russian journals covered by SCI in 1969 and 1972 are listed in the annual Guides. See the following: (a) Science Citation Index 1969 Guide Journal Lists. Philadelphia: Institute for Scientific Information, 1970; p. 12. (b) Science Citation Index 1972 Guide G Journal Lists. Philadelphia: Institute for Scientific Information, 1973; p. 105.
5. back to text Garfield E. Journal citation studies. II. Astrophysical Journal and its Supplements. Current Contents No. 35, 28 August 1974, p. 5-9.