Because there aren't any tag level clues available? The tags like preposition etc. are ambiguous.
In speech, prosodic means are often employed to designate phrasal boundaries. For example, final lengthening, intonation, pauses etc. mark the right (and some times the left) edge of boundaries.
but my question is for written language text. And the problem comes when I want to do phrase marking by machine
Then most possibly you need a representation of phrase structure as a whole; for that you will need first to normalise and tokenise the text, to tag it (possibly in POS), and then to parse the phrase. The parsing will provide the sentence structure; thus, you will be able to tell from that which are the phrase boundaries. See also this page: http://www.nltk.org/book/ch08.html, which provides an application in Python.
- Badges
- Science topic
- Similar topics
- Linguistics
- Philology
- Language