Comparing Online Scientific Info in Different Languages

Comparing Online Scientific Info in Different Languages

Being science literate is considered important; It is often thought as being able to find scientific information easily and use it when you need to, or when you want to.

However, many people struggle with these tasks for many reasons.

One reason is the “language divide": the disparities between those who know dominant languages such as English and those who do not.

So far, little is known about the availability of scientific information online in different languages.

Research Question

Are there differences in the quality of online scientific information between languages and scientific fields?


To investigate the research question, we undertook the following steps:

  1. We selected terms from three scientific fields (domains): Physics, chemistry and biology;
  2. We collected online search results regarding scientific terms in English, Hebrew, and Arabic;
  3. We analyzed their content, and rated their scientific and pedagogical quality.

The statistical methods included univariate and multivariate ANOVA and Linear Discriminant Analysis (LDA).


Findings indicate that searches in English yielded overall higher quality results, compared with Hebrew and Arabic, but mostly in pedagogical aspects, rather than scientific ones.
Overall, the differences in quality have more to do with the “language factor” than with the “scientific field factor” (Figure 1).
There appears to be a “language divide” in access to online science content. We conclude this because clustering the results by language yields better separation than clustering by scientific field (Figure 2).


We discovered that the quality of online scientific information differs across languages.

We think that scientific communities and institutions should take action to mitigate this problem. One way they could do this is by making high-quality information available in additional languages.


Figure 1. Effects of language and field on combined measures of information quality—MANOVA results. One, two, and three bullets (•, ••, and •••) denote statistical significance at the .05, .01, and .001 levels, respectively.


Figure 2. LDA of search terms (a) by language (colors) and (b) by field (shapes). (c) Differences in quality (areas of the triangles on the LDA plane) between equivalent terms in different languages. LD1 and LD2: First and second linear discriminants, respectively.


This study is first and foremost the brainchild of my colleague, Kawther Zoubi. She also carried out most of the study; I picked up where she left off. I conducted further analyses of the dataset and created the visualizations, re-did the literature review, and wrote up the manuscript.

The Linear Discriminant Analysis was conducted by my colleague, Dr. Eyal Nitzany.

Press Releases and Media Coverage


Zoubi, K., Sharon, A. J., Nitzany, E., & Baram-Tsabari, A. (2021). Science, maddá and ‘ilm: The language divide in scientific information available to internet users. Public Understanding of Science. Advance online publication.

Open Access