1

I'm looking for a colloquial English corpus that has time expressions marked in it. I don't need the time expressions to be coded in any special way, just that they be marked so I can find them. All of the corpora I've found are from news or broadcasting, and there's a big difference between the two for time expressions. I'm look for more complicated time expressions like "we meet every other Tuesday", etc.

4
  • In your example, 'we meet' is not a time expression. Only the adverbial 'every other Tuesday' is a time expression. The corpus would need to identify all time nouns, and further, identify all noun phrases containing those time nouns.
    – amI
    Jan 20 '16 at 18:05
  • @aml Sorry I wasn't clear; I need expressions such as "every other Tuesday", not "let's meet". The application is a calendar app with natural language understanding. I don't have an iPhone to test this, but if you say, "Siri, schedule my yoga class for every Wednesday at 2pm", the phone should add the yoga class to every Wednesday on your calendar at that time. Since we deal with natural language, we don't know what kind of weird time expressions people are going to try, so we need a corpus to test our coverage.
    – Nate Glenn
    Jan 21 '16 at 8:51
  • I can't point to a corpus, but I'll give you a hint: all adverbials of time would be prepositional phrases, with the null preposition '[at]' if not stated. Will you also have to deal with words like 'just', 'only', 'also', and 'too'?
    – amI
    Feb 3 '16 at 18:39
  • 1
    @aml We have to deal with natural language, meaning anything a large number of speakers can think of should be covered. Having a preposition is not guaranteed; "the meeting is every other Thursday", "Wednsday's appointment", etc. For now we've just had to contract out the corpus creation, but an open one would be a nice resource to have.
    – Nate Glenn
    Feb 4 '16 at 1:21

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.