3

I want to know what parts of speech (and in what frequency) follow a word or phrase. If I do a list search, for example "for the purpose of *", the result is a list of the actual words in order of frequeny. Is it possible to display the result as parts of speech instead? That is, instead of showing "private" (494 instances), followed by "this" (397), "the" (137), "making" (93), etc., it shows 1000 nouns, 500 verbs, 100 determiners, etc.

(1) I know there is a KWIC search, which color-codes the results according to their parts of speech, but the results seem to be (1) random and (2) arranaged alphabetically (as opposed to frequency.)

(2) ChatGPT gave me conflicting results. The first time, it told me to choose display as parts of speech from a dropdown menu in the results page, but I can't see such a choice in any dropdown menu. Could it be that I'm using the unlicensed version? The second time, it told me it was not possible.

7
  • Searching "for the purpose of" finds the words 'private' and 'this'?
    – Yosef Baskin
    Jan 29, 2023 at 21:30
  • This question seems more appropriate for meta.
    – user770884
    Jan 29, 2023 at 21:59
  • 3
    Don't trust ChatGPT: if you can't find something online, it won't be in ChatGPT. But there are other questions about COCA on Meta here.
    – Stuart F
    Jan 29, 2023 at 22:16
  • I’m voting to close this question because it is not about English language and usage. It is about software and a statistical exercise.
    – Anton
    Jan 29, 2023 at 23:16
  • 1
    @Anton First, corpora are directly related to studying usage. Second, if there anywhere the question is likely to get an answer, this would be the place. Third, the question is unsuited for any "software" or statistics stack exchange.
    – user468388
    Jan 29, 2023 at 23:27
  • @femke_0 Can you give a link to where ChatGPT tells you to "choose display as parts of speech from a dropdown menu in the results page". I can't find any such thing on their chat.openai.com page.
    – Mitch
    Jan 30, 2023 at 16:01
  • anything after "for the purpose of" will be a noun phrase of some sort.
    – Greybeard
    Jan 31, 2023 at 13:50

1 Answer 1

Reset to default
3

I don't think what you want is directly possible. You can, however, find this information somewhat manually.

Totals for one part of speech

Transform your wildcard into a specific part of speech, such as for the purpose of NOUN. The top right shows how many hits there are:

TOTAL 830
UNIQUE 483

Repeat for the parts of speech you want to count.

Finding popular parts of speech

As you can see below, under "Options", you can set the grouping to "None (Show PoS)":

group by: none

This returns results like this (with counts for each):

  • FOR (IF) THE (AT) PURPOSE (NN1) OF (10) THIS (DD1)
  • FOR (IF) THE (AT) PURPOSE (NN1) OF (10) THE (AT)
  • FOR (IF) THE (AT) PURPOSE (NN1) OF (I0) MAKING (VVG)
  • FOR (IF) THE (AT) PURPOSE (NN1) OF (I0) PROVIDING (WVG)
  • FOR (IF) THE (AT) PURPOSE (NN1) OF (10) OBTAINING (VVG)
  • FOR (IF) THE (AT) PURPOSE (NN1) OF (10) A (AT1)
  • FOR (IF) THE (AT) PURPOSE (NN1) OF (IO) CARRYING (VVG)
  • FOR (IF) THE (AT) PURPOSE (NN1) OF (IO) PREVENTING (VVG)
  • FOR (IF) THE (AT) PURPOSE (NN1) OF (I0) DETERMINING (VVG)

This doesn't seem terribly useful for your search, but it would if you didn't have a large number of unique hits. I found a list of tags in this BNC guide. (Remember, BNC is the same software running on a different dataset.)

You must log in to answer this question.