Using Ngrams

[Piscean]

As many members know, Ngrams are a very useful way of finding out about trends in the use of words and of comparing the relative frequency of expressions.

If, for example, you want to know whether people are generally bored of, by or with something, this ngram shows us that with is the most common preposition, and of the least common.

However, you need to be careful. This one shows a steep increase in the use of bored of since the 1970s, but it does not show that it is, at present, not as common as with or by. By changing the date settings, and comparing with alternatives, you get a less exaggerated picture of the steady increase in the use of of - 1985-2008 (the latest available date so far).

Do check the scale on the left of the graph. The 0,0000070 figure looks impressive in this graph, but less impressive in this one.

You can also compare the difference between BrE and AmE. There is no significant difference in the prepositions we use after bored in the two varieties, but there is a big difference with, for example, what we call our legal authority to drive in BrE and AmE (for some reason you'll have to click on 'Search lots of books' after you've followed the links to these two) .

Few of us bother to find out how to use ngrams effectively, so I have started this thread for people to give tips

My first tips, as I have noted above are:

- Check the scale on the left. An apparent increase/decline may not be as sharp as it seems.
- If you want to know how common an expression is, always check it with a similar expression on the same graph. (Use a comma to separate the two expressions.)
- Change to dates to ensure the graph is relevant to what you want to know.
- Don't forget to check BrE and AmE usage. The results may surprise you.

Here are a couple more:

- Click the case-insensitive box unless capitalisation of a particular word is important.
- If you type _INF after the base form of a verb, you'll get a graph of all forms of the verb, as here

If you have other tips, please post them in this thread.

[Tdol]

You can also find Ngrams and other databases here:

http://corpus.byu.edu/

[Piscean]

I have just discovered something new. If you add _VERB or _NOUN to the word you enter, the the forms will appear separately. The following two Ngrams show how schedule as verb became more common in AmE than in BrE. Note that the scales on the two graphs are different. Both noun and verb are more commonly used in AmE than in BrE.

[Mori]

See About Ngram Viewer for more tips and tricks.

[nyggus]

Hi,

I love Ngrams and use them a lot. But I do have problems with understanding the percentages. You might be tempted to interpret them as shares of the corpus sources in which the phrase occurs. Well, this would be a natural and easy and great intepretation, wouldn't it? But...

Look here... Either the percentages should be interpreted differently or something is wrong with the way they are calculated. I don't believe Google would made such a mistake, so the shares should be interpreted in some other way. Can anyone hint how?

Thanks,
Nyggus

[GoesStation]

Like you, I had guessed that the percentages represented the number of books in which the subject text occurred. Your interesting example leads me to a simpler hypothesis: they're the share of words in the corpus. In other words, 5.2% of all words in the corpus are "the".

[nyggus]

Hey, it makes sense! If we add some prepositions, we get something like this.

Another thing is that when explaining how Ngrams Viewer works, Google might say that directly.

Other than that, the Ngrams guide shows powerful ways of using this tool. I remember I was rather reluctant to read it (at first glance, the guide had not seemed very encouraging to me), but when I actually did read it, Ngrams repaid my time and effort. I think intermediate and advanced learners (not necessarily beginners, though) can gain a lot from using Ngrams, but they do have to remember to use it with some care.

[Tdol]

Ahh, the benefits of reading the manual. Something that very few do.

[GoesStation]

Ahh, the benefits of reading the manual. Something that very few do.

Like asking for directions, reading manuals makes men lose their hair.

[nyggus]

Like asking for directions, reading manuals makes men lose their hair.

Let me look around. I've always been wondering who among my male colleagues has read too many manuals. Now I know how to check. (Alas, I myself show signs of having spent too much time with manuals...)

What about women?