This question was inspired by the comment/chat discussion under this answer. This would be very helpful to know.

  • 2
    without weighing in on the substantive reasons, I would note that BNC only contains text from 1992 and earlier (and so is basically 20 years out of date)
    – nohat Mod
    Commented Jul 18, 2011 at 23:51
  • @nohat The Collins corpus is up-to-date but it costs £695 a year for 10 users. :(
    – z7sg Ѫ
    Commented Jul 19, 2011 at 11:01

1 Answer 1


The problem with using the 'British English' corpus selection on ngrams is that the corpus includes American books that were published in the UK. It is not a selection of British texts, so it does not accurately represent English as used in the UK.

This ngram for fantasize vs. fantasise is misleading. UK spelling convention is fantasise despite the ngram and whatever was written in the original question thread.

bad ngram

The BNC (British National Corpus) is a 100 million word collection of samples of written and spoken language from a wide range of sources, designed to represent a wide cross-section of British English.

Excluding spoken English, a search of the BNC revealed 34 counts of fantasise, but only 16 for fantasize, of which 8 are from the American magazine Esquire. Regardless of dates, this clearly disagrees with the ngram.

The 'British English' google corpus can still be useful in comparing relative usage frequencies between British and American English over time, but care should be taken when interpreting the results, and I certainly don't think it should be relied on to settle matters of British English usage.

  • +1, very informative; could you also provide a link or references
    – Unreason
    Commented Jul 19, 2011 at 9:44
  • The -ise suffixes (apart from analyse) seem to behave quite a bit differently than other British/American spelling differences. According to Google Ngrams, for all of these ise words I've checked, ise appears to be the preferred spelling only between ca. 1870 and 1930, whereas Google Ngrams does a very creditable job on spelling differences like color/colour, etc. There is something funny going on with the ise words. I don't know what. Commented Jul 20, 2011 at 13:13
  • @Peter Well I'm not sure it does a creditable job with colour/color either. BNC returns 11344:115 because color is regarded as a US-only spelling. If you were to interpret the ngram for the pairing, you might draw the conclusion that colour is simply the favoured variant. That would be incorrect. The bump at the end interests me though. If only I had access to a more modern corpus. (I hope if I mention this enough times, the gods will grant it to me somehow!)
    – z7sg Ѫ
    Commented Jul 20, 2011 at 13:24
  • @z7sg: I hope you will agree that it does a lot better for colour/color than for any of the ise words. The "color"s that slipped through probably are misclassifications, but that level of misclassification (I estimate less then 10%) is too small to account for the ise Ngrams. I am wondering if maybe the type of document has something to do with it; for example, if ize was more common in science writing, this might account for the difference between the corpora. Commented Jul 20, 2011 at 13:39
  • @Peter Assuming that the proportion of color vs colour indicates the proportion of American vs British sources present in the corpus, then yes I would agree that there must be some other factor contributing to the high frequency of -ize words.
    – z7sg Ѫ
    Commented Jul 20, 2011 at 14:54
  • 1
    @Peter, Oxford University Press prefers -ize spellings, which are generally (with, I think, about six exceptions, one of which is advertise) considered acceptible in British English, though the -ise versions are more common. ISO standards use British spellings with the -ize variants.
    – TRiG
    Commented Jul 21, 2011 at 14:44

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .