请问怎样利用在线的母语语料库与CLEC进行overuse和underuse的比较

[pepsi_blue]

我手头只有CLEC的语料库,可是想与本族语者的某些连接词进行比较,看哪些是overuse和underuse.但是在线的母语语料库能进行keywords的查询吗?如果有的话,哪些可以呢?

 

[xiaoz]

回复: 请问怎样利用在线的母语语料库与CLEC进行overuse和underuse的比较

You actually do not need to do keyword analysis to find out about under- or overuse. Frequencies are sufficient for this.

 

[pepsi_blue]

回复: 请问怎样利用在线的母语语料库与CLEC进行overuse和underuse的比较

ok, thank you Mr. Xiaoz.

 

[xujiajin]

回复: 请问怎样利用在线的母语语料库与CLEC进行overuse和underuse的比较

Once you get the frequency, together with the corpus size, you can use chi-square or log likelihood test for overuse and underuse.

 

[pepsi_blue]

回复: 请问怎样利用在线的母语语料库与CLEC进行overuse和underuse的比较

thank you xu, but i still have two questions: 1.can i use antconc to test chi-square or log likelihood, (i know chi-square can be tested by SPSS, do you mean that if i test chi-square, i have to copy the data to the SPSS)
2.what is the difference between chi-square and log likelihood

 

[xujiajin]

回复: 请问怎样利用在线的母语语料库与CLEC进行overuse和underuse的比较

Keywords analysis in AntConc use log likelihood as its default signigicance test method (chi-square test is also available).

Fequency counts should be exported to SPSS to perform the tests.

Both chi-square and log likelihood tests will do for your purposes. But according to statistician, likelihood ratio yields improved result.
(In nature, log likelihood is an improved type of chi-square test.)

 

[pepsi_blue]

回复: 请问怎样利用在线的母语语料库与CLEC进行overuse和underuse的比较

Thank you for patient explanations, Dr. xu. That is to say, i have to use both SPSS and AntConc for my statistics. The former is to get the frequency and the latter for the log likelihood test.

 

[xujiajin]

回复: 请问怎样利用在线的母语语料库与CLEC进行overuse和underuse的比较

Yes. It is preferrably so.