Often when you look for some new word in a dictionary, you will find many definitions or meanings of the one word.

For example when you look for a word to in Oxford Advanced Learner's Dictionary you will find the sorted list of 18 possible meanings of the word.

The easier part of the question: The words are sorted according to what? The word frequencies? Is meaning number 1 is most often used?

The harder part of the question: Is there a dictionary evaluating each item of the list (of meanings) with percentage?

For example for the word to it would look like:

  1. in the direction of something; towards something (in 60 % of meanings)

  2. to the something (of something) located in the direction mentioned from something (in 22 % of meanings)

  3. as far as something (in 10 % of meanings)

  4. ... (in 5 % of meanings)

  5. ... (in 1 % of meanings)

  6. ... (in 0.3 % of meanings)


18 ... (in 0.0001 of meanings)

If such a dictionary does not exist and you know about some research about this, you can cite such a source.

  • 1
    Can you define what you mean by 'x% of meanings'? If it's "Number of times each has been used somewhere in English" no such source can ever exist, not least because the numbers are constantly changing. Commented Dec 9, 2011 at 15:23
  • 2
    @TimLymington The frequency (percentage) could be taken from some corpus or their combination, but such a corpus should also take into account context to determine the correct meaning (sense, not only part of speech). Or course such a dictionary have to cite the corpus according to what the frequencies are. I know only corpora that don't take meaning (senses) into account.
    – xralf
    Commented Dec 9, 2011 at 15:33
  • There is always the issue of current usage when we base our statistics on the corpora. If we are only interested in the cumulative historical application of the term, then it's fine.
    – Kris
    Commented Dec 11, 2011 at 10:07

2 Answers 2


The number of senses a dictionary reports depend a good deal on how much space it has and how long the lexicographers are allowed to parse the data. Typically, the more they look, the more they find. The Longman Dictionary of Contemporary English uses corpus data to sort its senses by frequency though they don't provide the sense frequency information. Interestingly, Adam Kilgarriff, who worked on the 3rd edition of the LDOCE points out that in almost all cases, the most frequent sense of a word is overwhelmingly so. Adam's website with a number of relevant papers is here.

  • Thank you. I tried online version and it's good dictionary. You could cite the source where is written that the senses in LDOCE are sorted yet. I found one interesting research that could complement this answer with Affable Geek's answer. Maybe percentage information could be added later, if they have numbers for sorting.
    – xralf
    Commented Dec 15, 2011 at 10:50
  • 1
    The dictionary itself is the source. In the 2003 edition on p. xi it says, "The meanings in the entries are as afar as possible ordered in accordance with their frequency in the language as shown by the Corpus, which means that the most frequent meanings are placed first." Commented Dec 15, 2011 at 12:19

It really depends on the dictionary as to what order they are. Some group meanings logically, others chronologically, and most according to how "common" they are - but that commonness is usually subjective.

If you want word frequencies, you need a corpus, not a dictionary. Look at Google Books search, for example, if you want to know how frequently a word is used in a given phrase or part of speech. in theology, trying to get the meaning of a Greek word as it was used at the time the Bible was written, there are corpses of "Greek works at the time of the Bible". You'd want to differentiate that from, say, modern Greek, for the same reason you'd want to differentiate, say, slang from scientific research.

You need a corpus, because you need to define your total population of words, the denominator of your percentage. There simply too many words to get a frequency otherwise - even Google books, one of the largest publically accessible corpuses, indexes less than 10% of the books out there.

For more information on corpuses, you should check out Grammar Girl's podcast which recently did an episode on one by clicking here.

Unfortunately, this is going to be able to show you how a word is used, but semantic meaning is really tricky. You're best bet would be to isolate meaning by how its used. Since there is no standard on the various meanings of a word, you'd have to find a dictionary tied to a particular corpus. For theological work, you can often find tagged grammars that might get you what you want - but if you're looking for a more general purpose dictionary, I strongly suspect you'll be out of luck. Since knowing the exact percentage of a given use is such a specialized query, I doubt its worth anyone's time to compile an exhaustive list of several words tied to an arbitrary set of meanings. If you can narrow it to a single word (or several words), you can probably do your own, but I'd be amazed if the resource you want exists.

  • 2
    I think that most dictionaries use some combination of these sorting criteria. I also believe that some dictionaries also sort the meanings according to which was used first chronologically. Commented Dec 9, 2011 at 14:54
  • @Affable Geek I'm aware of the fact that the dictionary would use corpuses. But my question is not only about frequencies, but frequencies of meanings and that's the difference. The corpus should take into account contextual information from the corpus and the dictionary should utilize corpus that works with contextual information. I'm also aware of the fact that there are many corpuses.
    – xralf
    Commented Dec 9, 2011 at 15:06

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.