Loading corpus…
0
No text held
Thin week (under 300 words)
Key event
Pick a country. Each row is a recurring formula from that government's official English record, and each column is a week (Monday to Sunday). By default a cell shows how often the phrase was used per 1,000 words of official text that week, so a week with one long speech and a week with ten short statements are on the same footing. The other measure, share of statements, counts the transcripts or statements that used the phrase at least once.
Click a cell to read the sentences behind it in the panel, with the phrase highlighted and a link to the official page. Arrow keys move the selection once the heatmap has focus; clicking a row label jumps to that phrase's busiest week. Grey cells mean no text is held that week. Hatched cells are thin weeks with under 300 words, where one sentence can swing the rate. The strip under the axis shows how many words the corpus holds each week.
Labels above the grid are key events, each with a source in the list below. Click one to move the selection to its week and compare the rhetoric before and after.
scripts/collect_kremlin.py, which honours the site's robots.txt and spaces its requests.scripts/collect_iran_mfa.py.scripts/build_data.py from the dictionary in scripts/dictionary.py. Key-event sources are listed with each event.Every phrase on the heatmap, its pattern (a case-insensitive regular expression; \b marks a word boundary, \w* any word ending), how many items use it, and the earliest and latest uses in the corpus.