The paper argues for greater linguistic sensitivity in scientometric research, discussesBaldauf andJernudd,11–13 and raises issues of North/South, English/non-English imbalance in research communication. It then proposes a procedure for identifying native/non-native speaker status of authors in English periodical literature on textual evidence. Preliminary application to 623 articles is reported and evaluated. The Health Science NNS percentage was 23%; in Economics half that. In both there were few papers of Third World provenance, thus supportingBaldauf andJernudd. It is suggested that Scientometrics could contribute to the teaching of Research English, against a background of adjusting suspected imbalance.