I would like to know where can I find statistics like the one on the picture. It comes from this answer.
asked Sep 4, 2019 at 22:37
Add a comment
|
2 Answers
The answers to the question linked to by this answer don't mention it, but a good alternative to the Google Ngram Viewer is COCA which, although smaller by 3 orders of magnitude, does allow for context inspection, which can be used to verify the numbers (false positives always creep their way into the results of a query). COCA is restricted to AmE, but if you're interested in BrE in particular, use BNC, which is 5 times smaller and isn't quite as updated. When using the Google Ngram Viewer, don't go beyond the default range (1800–2000).
-
2I'd go one step further and mention how sometimes it's useful to restrict the dates, and not necessarily use the entire default range. Some ngrams don't need two centuries worth of information to make their point (like this one), and restricting the dates makes the ngram easier to read and analyze.– J.R. ModCommented Sep 6, 2019 at 12:52
The images are from Google’s Ngram Viewer.
It’s a useful tool, but the results can sometimes be misleading because they don’t take context into account. The What is Ngram and what is it used for? discussion has more information about them.
-
There's also the case of very uncommon words or phrases in which Ngrams are from downright buggy and nonsensical to useless because of too few data points.– M.A.R.Commented Sep 7, 2019 at 10:48
-
@M.A.R. True, ngrams aren't very good for modern slang either. I often wonder about the usefulness of a corpus that consisted entirely of dialogue from movie scripts and plays.– ColleenVCommented Sep 7, 2019 at 10:56