Friday, September 01, 2017

How Often Do European Language Journals Cite English-Language vs Same-Language Work?

By Eric Schwitzgebel and Ivan Gonzalez-Cabrera

Elite English-language philosophy journals cite almost exclusively English-language sources, while elite Chinese-language philosophy journals cite from a range of linguistic traditions.

How about other European-language journals? To what extent do articles in languages like French, German, and Spanish cite works originally written in the same language vs. works originally written in other languages?

To examine this question, we looked at a convenience sample of established journals that publish primarily or exclusively in European languages other than English -- journals catalogued in the Philosophy section of JStor with available records running at least from 1999 through 2010. [note 1] We downloaded the most recently available JStor archived issue of each of these journals and examined the references of every research article in those issues (excluding reviews, discussion notes, editors' introductions, etc.). This gave us a total of 96 articles to examine, 41 in French, 23 in German, 14 in Italian, 8 in Portuguese, 6 in Spanish, and 4 in Polish.

Although this is not a systematic or proportionate sample of non-English European-language journal articles, we believe it is broad and representative enough to provide a preliminary test of our hypothesis. Are citation patterns in these journals broadly similar to the citation patterns of elite Anglophone journals (where 97% of citations are to same-language sources)? Or are they closer to the patterns of elite Chinese-language journals (51% of citations to same-language sources)?

In all, we had 2883 citations for analysis. For each citation, we noted the language of the citing article, whether the cited source had originally been published in the same language as the citing article or in a different language, and if it was a different language whether that language was English. As in our previous studies, sources in translation were coded based on the original language of publication rather than the language into which it had been translated (e.g., a translation of Plato into German would be coded as ancient Greek rather than German). We also noted the original year of publication of the cited source, sorting into one of four categories: ancient to 1849, 1850 to 1945, 1946-1999, or 2000-present. [note 2]

In our sample, 44% of citations (1270/2883) were to same-language sources, 30% were to sources originally published in English (some translated into the language of the citing article), and 26% (749/2883) were to all other languages combined. These results are much closer to the Chinese-language pattern of drawing broadly from a variety of language traditions than they are to the English-language pattern of citing almost exclusively from the same linguistic tradition.

French- and German-language articles showed more same-language citation than did articles in other languages (51% and 71% respectively, compared to an average of 20% for the other sampled languages), but we interpret this result cautiously due to the small and possibly unrepresentative samples of articles in each language.

Breaking the results down by year category, we found the following: [if blurry, click for clearer display]

Thus, in this sample, cited sources originally published between 1946 and 1999 were just about as likely to have been originally written in English as to have been written in the language of the citing article. When the cited source was published before 1946 or after 1999, it was less likely to be in English.

Looking article by article, we found that only 5% of articles (5/96) cited exclusively same-language sources. This contrasts sharply with our study of articles in Anglophone journals, 73% of which cited exclusively English-language sources.

We conclude that non-English European-language philosophy articles cite work from a broad range of linguistic traditions, unlike articles in elite Anglophone philosophy journals, which cite almost exclusively from English-language sources.

One weakness of this research design is the unsystematic sampling of journals and languages. Therefore, we hope to follow up with at least one more study, focused on a more carefully chosen set of journals from a single European language. Stay tuned!


note 1: Included journals were Archives de Philosophie, Archiv für Rechts- und Sozialphilosophie, Crítica: Revista Hispanoamericana de Filosofía, Gregorianum, Jahrbuch für Recht und Ethik, Les Études Philosophiques, Revista Portuguesa de Filosofia, Revue de Métaphysique et de Morale, Revue de Philosophie Ancienne, Revue Internationale de Philosophie, Revue Philosophique de la France et de l'Étranger, Rivista di Filosofia Neo-Scolastica, Rivista di Storia della Filosofia, Roczniki Filozoficzne, Rue Descartes, Sartre Studies International, Studi Kantiani, and Studia Leibnitiana. We excluded journals for which substantially more than half of recent articles were in English, as well as journals not listed as philosophy journals on the PhilPapers journals list.

note 2: Coding was done by two expert coders, each with a PhD in philosophy. One coder was fluent only in English but had some reading knowledge of German, French, and Spanish. The other coder was fluent in Spanish and English, had excellent reading knowledge of German and Portuguese, and had some reading knowledge of French and Italian. The coding task was somewhat difficult, especially for journals using footnote format. Expertise was required to recognize, for example, the original language and publication period of translated works, which was not always immediately evident from the citation information. We randomly selected 10 articles to code for inter-rater reliability, and in 91% of cases (235 of 258 citations) the coders agreed on both the original language and the year-category of original publication. Errors involved missing or double-counting some footnoted citations, typographical error, or mistakes in language or year category. Errors did not fall into any notable pattern, and in our view are within an acceptable rate given the difficulty of the coding task and the nature of our hypothesis.

No comments: