5

I'm new to Corpus Linguistics and I'm writing a paper about the English and Portuguese "because noun", a type of construction such as "I'm going home because GTA5".

However, when I try to search this type of structure on corpora, I find mainly VPs, such as "I'm going home because GTA5 is a great game".

Is it possible to filter my search results with because + NP + punctuation mark (period or exclamation mark)? This would avoid undesirable results containing whole VPs.

enter image description here

Improve this question
4
  • 1
    Sure, this is usually done for amongst others the reasons you mentioned. What the exact tag names for punctuation marks are of course depends on the individual tagger that you use. Which is the one in your screenshot?
    – Natalie Clarius
    Oct 11, 2016 at 16:20
  • @lemontree Sorry, I don't know what exactly the tagger is. I'm using the "Corpus of Global Web-Based English" corpus.byu.edu/glowbe Trying to work with the following string: "because [j*]", but it's too troublesome without the ponctuation marks.
    – Matt
    Oct 12, 2016 at 0:50
  • The tagger is the algorithm that assigns the tags. Different taggers may use different tag sets (e.g., some more elaborate taggers provide different tags for different verb forms, like indicative or participle forms, more simple ones would subsume everything under "verb"), therefore it is crucial for answering the question what tagger/tag set is being used. Looks like what you're looking for is PUNC for "punctuation" (last entry in the drop-down list).
    – Natalie Clarius
    Oct 12, 2016 at 16:23
  • Because of what Lefty G Balogh wrote in his answer, I'm afraid you probably won't be able to do your search with this tool: The parser won't accept two POS tags without a space in between, but obviously inserting whitespace between the last word in a sentence and the punctiation mark won't yield any results, so apparently you can't query for because [n*][y*]. Try another corpus instead, e.g. the web and chat text corpus that comes with Python NLTK, a great NLP library which provides various tools for POS extraction as well.
    – Natalie Clarius
    Oct 16, 2016 at 16:03

2 Answers 2

Reset to default
1

I'd say if you are trying to limit yourself to noun phrases, you are probably better off with "because of NP". I'd suggest you use the following in the search bar, and limit yourself based on the resulting table:

because of [n*]

The tagger, as pointed out above seems glitchy as it does not interpret the punctuation tag properly. This because of [n*][y*] fails badly, and I could not squeeze out a reasonable output with any variation at all.

PUNC would be the punctuation mark, but it does not seems to parse correctly: enter image description here

Improve this answer
3
  • 1
    I think that the OP was in fact interested in because [n*]. rather than because of [n*], where inserting punctuation is crucial so as to exclude constructions like because [n*] was a ... - see his other question. Anyway, yes, [n*][y*] seems to fail: The parser won't accept two POS tags without a space in between, but obviously inserting whitespace between the last word in a sentence and the punctiation mark won't yield any results.
    – Natalie Clarius
    Oct 16, 2016 at 15:03
  • Roger that. Out of context, I automatically corrected it. But even if I take that into consideration, I agree with your response to that other question. I'd probably even be hesitant to use social media as a reliable corpus.
    – Lefty G Balogh
    Oct 16, 2016 at 15:45
  • 1
    Thanks for the help, guys! I'm having success so far with the string [code]because ADJ .[/code] on COCA. I haven't tried any other POS though, but I guess they should work the same way.
    – Matt
    Oct 16, 2016 at 18:31
1

In my experience, it is pretty standard in corpus linguistics to include punctuation as tokens and give them appropriate part-of-speech tags, e.g. in the UPENN tagset, there is a tag SENT for the full stop closing a sentence.

Note that automatic tokenisers and POS taggers aren't infallible, in your scenario you may encounter false positives because abbreviations unknown to the tools come out as spurious end-of-sentence.

Improve this answer

Your Answer

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

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