Dataset Information

Automated content analysis across six languages.

ABSTRACT: Corpus selection bias in international relations research presents an epistemological problem: How do we know what we know? Most social science research in the field of text analytics relies on English language corpora, biasing our ability to understand international phenomena. To address the issue of corpus selection bias, we introduce results that suggest that machine translation may be used to address non-English sources. We use human translation and machine translation (Google Translate) on a collection of aligned sentences from United Nations documents extracted from the Multi-UN corpus, analyzed with a "bag of words" analysis tool, Linguistic Inquiry Word Count (LIWC). Overall, the LIWC indices proved relatively stable across machine and human translated sentences. We find that while there are statistically significant differences between the original and translated documents, the effect sizes are relatively small, especially when looking at psychological processes.

SUBMITTER: Windsor LC

PROVIDER: S-EPMC6867602 | biostudies-literature | 2019

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

Automated content analysis across six languages.

Windsor Leah Cathryn LC Cupit James Grayson JG Windsor Alistair James AJ

PloS one 20191120 11

Corpus selection bias in international relations research presents an epistemological problem: How do we know what we know? Most social science research in the field of text analytics relies on English language corpora, biasing our ability to understand international phenomena. To address the issue of corpus selection bias, we introduce results that suggest that machine translation may be used to address non-English sources. We use human translation and machine translation (Google Translate) on ...[more]

PMID: 31747404

Similar Datasets

Project description:BackgroundAlthough subjective expressions and linguistic fluency have been shown as important factors in processing and interpreting textual facts, analyses of these traits in textual health information for different audiences are lacking. We analyzed the readability and linguistic psychological and emotional characteristics of different textual summary formats of Cochrane systematic reviews.MethodsWe performed a multitrait-multimethod cross-sectional study of Press releases available at Cochrane web site (n = 162) and corresponding Scientific abstracts (n = 158), Cochrane Clinical Answers (n = 35) and Plain language summaries in English (n = 156), French (n = 101), German (n = 41) and Croatian (n = 156). We used SMOG index to assess text readability of all text formats, and natural language processing tools (IBM Watson Tone Analyzer, Stanford NLP Sentiment Analysis and Linguistic Inquiry and Word Count) to examine the affective states and subjective information in texts of Scientific abstracts, Plain language summaries and Press releases.ResultsAll text formats had low readability, with SMOG index ranging from a median of 15.6 (95% confidence interval (CI) 15.3-15.9) for Scientific abstracts to 14.7 (95% CI 14.4-15.0) for Plain language summaries. In all text formats, "Sadness" was the most dominantly perceived emotional tone and the style of writing was perceived as "Analytical" and "Tentative". At the psychological level, all text formats exhibited the predominant "Openness" tone, and Press releases scored higher on the scales of "Conscientiousness", "Agreeableness" and "Emotional range". Press releases had significantly higher scores than Scientific abstracts and Plain language summaries on the dimensions of "Clout", and "Emotional tone".ConclusionsAlthough the readability of Plain language summaries was higher than that of text formats targeting more expert audiences, the required literacy was much higher than the recommended US 6th grade level. The language of Press releases was generally more engaging than that of Scientific abstracts and Plain language summaries, which are written by the authors of systematic reviews. Preparation of textual summaries about health evidence for different audiences should take into account readers' subjective experiences to encourage cognitive processing and reaction to the provided information.

Dataset Information

Automated content analysis across six languages.

Publications

Automated content analysis across six languages.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets