[Corpora-List] what's multi-word unit

[Piao, Songlin]

Hi,

It has been quite a few years since people started reporting works on Multi-Word Unit (MWU) extraction, but seems like there is not clear definition of MWU yet. In some cases, we can extract many grammatical word strings like "come from", "live with" but are they MWUs? If not why?

I personally prefer to think MWUs as relatively stable word groups (mostly adjacent ones) that have relatively highly co-occuring probabilities than free conbinations, but that may have more flexible structures than dioms or fixed phrases do. Am I right, or is there any "standard" definition of MWU?

Any suggestions and comments will be appreciated.

Scott Piao
----------------------------
Dept. of Linguistics and MEL
Lancaster University
Lancaster LA1 4YT
UK