Google spell check

The fairly reliable practice of using Google to check one's spelling. Not only will Google usually suggest the correct spelling, but it will let you get an idea of how prevalent one's own misspelling is. This gives rise to the Google Common Use Index (GCUI), the ratio of appearances of a particular incorrect spelling versus the correct spelling.

Some examples:

• handfull vs. handful: 26,500/901,000 = 2.9%
• definitly vs definitely: 53,300/3,360,000 = 1.6%
• irregardless vs. regardless: 17,200/3,410,000 = 0.5%

• Georgia O'Keefe vs. Georgia O'Keeffe: 22,500/30,100 = 74.8%
• supercede vs. supersede: 36,200/163,000 = 22.2% (Google will not suggest the correct spelling)

The GCU ratio has many potential uses, such as the following Chatterbox moment. Robinspoon suggested that spon might be more common (versus spoon) in Canada than other parts of the word. (Why this was proposed is fortunately irrelevant.) Using the versitile tool that is the GCUI, RS was found to be 'correct':

• spon vs. spoon, globally: 77,900/752,000 = 10.35%
• spon vs. spoon, in Canada: 1,480/12,500 = 11.84% ("site:*.ca" appended to query).

There is a certain irony here: Google is itself a misspelling of googol, a mathematical term for 10¹⁰⁰ (one followed by 100 zeros). Thus, the string of 'o's at the bottom of the Google search results page. Google's GCUI is a whopping 24,720% — 3,090,000/12,500!

nodeshell rescue

Although its user interface has become a nightmare by comparison with Google's, it should be noted that you may be better off with AltaVista when the aim is more to check on term usage than to get a spellcheck suggestion per se. As AV ORs the search terms entered by default (whereas Google ANDs them as well as ignoring stopwords even in quoted strings), you can enter a range of variants of the proposed word in one easy operation, and, near the foot of the results page, you will find:

The number of words that match your search terms: on line banking 25017 ? online banking 600771

Nonetheless, the fact that Google's results page shows the term in context, as well as its rather larger database and access to file formats other than HTML, means that Google is still at present the better general tool for terminological research.

In both Google and Alta Vista, searches can be restricted by domain: in AV use domain:.uk for Britspeak; to restrict your hits to American English the best bet is to use site:.edu (Google) or domain:.edu (AV) since the (unlike .com or .net) the use of the .edu top level domain is generally restricted to the USA (I guess that .mil would also work - depends on your subject area, I guess); however note that in this case AV's usage figures are the total number of hits from all domains, not just the one you asked for.

koreykruse points out that Google offers sundry boolean alternatives under "advanced search"; it does indeed, but (a) it doesn't handle quotes very well which makes it unusable for different versions of multi-word terms and (b) it doesn't break down the number of hits for each alternative, just gives you a total. gn0sis observes, meanwhile that the AV interface is less messy and lower bandwidth if you go via the text-only interface available variously via:

• http://www.altavista.com/sites/search/text
• http://www.raging.com
• http://www.ragingsearch.com

(Raging was originally a short-lived Google-look-and-feel interface to the same AltaVista database but has now been redirected to the old AV text interface).