This is a matrix of graphs showing how many occurences there are of 'rocher' each year (normalised per million words) in each genre of text on the corpus, in each dialect. This should give a visual indication of which dialects use the word more frequently and which genre of text use the word more frequently.
A couple wee notes to bear in mind: Not all genres are well populated, there are very few 'scripts', the 'social' genre representing social media only really got going in 2010, some dialects have no newspaper text represented in the corpus. Also some dialects are sparsely populated, Orkney and Southern have fewer than 100,000 words and so the coverage by year is patchy.