[求助]请教高手yinghuang!谢谢!

turf2008

初级会员
Verb ST4 ST6 COCNUS
do 650 230 123
get 256 182 86
give 154 120 83
make 206 157 147
have 493 419 400
take 151 151 129
(每200000词中的频数)
作者用了SPSS中的卡方检验,显著水平为5%。检验的结果如下:
verb ST4 vs COCNUS ST6 vs COCNUS ST4 vs ST6
do 283.02 27.83 212.56
get 70.78 29.65 13.76
have 8.67 0.39 6.69
take 60.21 1.66 56.38
give 18.17 5.79 4.73
make 7.21 0.30 17.92
没有过程,我不懂作者是怎么算的,我也想用这个方法,你帮我看看第二个表格中的结果是如何算出来的吧。谢谢!
 
我以make的频数比较了在ST4和COCNUS中使用的情况。根本得不出chi-square=7.21这个值。看来你只能问原作者了。
 
回复:[求助]请教高手yinghuang!谢谢!

不好意思,能告诉我怎么做吗?谢谢!谢谢!
 
回复:[求助]请教高手yinghuang!谢谢!

一个比较简单的方法就是卡方(拟合)检验。仍以make在上述两语料中的频数为例,直接将206和147输入spss,并定义或者说设定分类变量。然后到data工具按钮选weight cases,选中weight cases by前的复选框,将frequency变量选入。然后到Analyze按钮选nonparametric tests: chi-square.将分类变量选入,其余默认。ok即可。

[本贴已被 作者 于 2006年07月28日 12时17分14秒 编辑过]
 
Observed N Expected N Residual
147.00 147 176.5 -29.5
206.00 206 176.5 29.5
Total 353
Test Statistics

a
Chi-Square(a) 9.861
df 1
Asymp. Sig. .002
a 0 cells (.0%) have expected frequencies less than 5. The minimum expected cell frequency is 176.5.

是这样的结果吗?
 
I think the author might have used the following as the basis for computing chi-square (but I am not sure):

word number of ST4
raw frequency of "make" in ST4
word number of COCNUS
raw frequency of "make" in COCNUS

and this is the correct method (but I don't know it that computation will lead to 7.21).
 
回复:[求助]请教高手yinghuang!谢谢!

Perhaps Dr.Xiao is correct. But we don't know the exact total numbers of running words in ST4 and COCNUS respectively. The formula Dr. Xiao proposes is the corrected continuity chi-square as follows:
2006072818583014.gif



以下是引用 xiaoz2006-7-28 17:30:23 的发言:
I think the author might have used the following as the basis for computing chi-square (but I am not sure):

word number of ST4
raw frequency of "make" in ST4
word number of COCNUS
raw frequency of "make" in COCNUS

and this is the correct method (but I don't know it that computation will lead to 7.21).
 
回复:[求助]请教高手yinghuang!谢谢!

如果我们比较Brown和LOB中make的频数的差异显著性,原始数据如下:
Make in Brown:24(公式中的a)
Make in LOB:41(公式中的b)
Total no. of words in Brown: 1,000,000 (公式中的c)
Total no. of words in LOB: 1,000,000 (公式中的d)

n=a+b+c+d
v(自由度)=1


则spss使用上述公式计算的chi-square 如下表:
2006072820210148.gif



[本贴已被 作者 于 2006年07月28日 20时37分36秒 编辑过]
 
Verb ST4 ST6 COCNUS
do 650 230 123
get 256 182 86
give 154 120 83
make 206 157 147
have 493 419 400
take 151 151 129
(每200000词中的频数)
以上为normalized frequency,而卡方检验结果是基于原始数据,该种处理方法源于Altenberg & Granger (2001);若要原始数据请和我联系。
 
Back
顶部