A Four-Month Look At Part Of Speech Distribution In A Multilingual World

In January, we visualized one week's worth of the part of speech distribution of the 11 languages annotated by Google's Natural Language API that make up GDELT's Web Part Of Speech dataset:  Chinese (Simplified), Chinese (Traditional), English, French, German, Italian, Japanese, Korean, Portuguese (Brazilian & Continental), Russian and Spanish. Looking across these languages, what does the overall part of speech distribution look like in such a multilingual world?

The hierarchical chart below shows the overall distribution across the 11 languages over the period January 17, 2020 through present. Despite covering a third of a year of global news coverage, compiling this graph took a single SQL query in BigQuery just 3 seconds to create!


Compiling the chart above took just a single line of SQL:

select lang, posTag, sum(count) count FROM `gdelt-bq.gdeltv2.web_pos` where DATE(dateTime) >= "2020-01-17" group by lang, posTag order by lang asc, posTag asc