紧急 求助!双宾构式的检索

因研究需要,我想用语料库检索动词出现在双宾和介词宾语构式中的频率。目前在使用Brigham Young University Interface的BNC和Contemporary American English Corpus。因本人完全是个生手,所以检索的过程中碰到一些问题。特向各位高手求助。
1. The corpora: BNC and Corpus of Contemporary American English Corpus

2. Purpose: To count the frequencies with which the dative verbs like give are used with the double object (DO) like John gave Mary a book/ John give me (her) a book and prepositional dative construction like John gave a book to Mary (PD).

3. The syntax I used:
DO (Double object constrcution):
For constructions with lexical NP as Recipient: give [nn*] [nn*] context:0/1
For constrcutions with pronouns as Recipient: give [pn*] [nn*]

PD (Prepositioanl dative construction):
give [nn*] to context:0/1

4. Puzzles:
1) When counting the frequencies of the ditransitive constrcution, phrases like “give expert advice”, “give health care” cannot be excluded from the search results. Regarding the prepositional dative construction, idioms like “give rise to” have to be deleted manually. Is there any way that phrases like these can be ignored by defining the syntax?

2) Only two pronouns used as recipients after give were found when I used the search expression give [pn*] [nn*], that is, someone and anyone and there are only 25 instances in total. Since I doubted the research results, I conducted a new search with the syntax give me [nn*], I got 1123 double object constructions altogether, suggesting the syntax I used above is not effective. I’m wondering whether there is a way that all the double object constructions with pronoun reciepients could be obtained at one time.
Your kind help would be greatly appreciated!
 
回复: 紧急 求助!

可以试试用 regular expression 来解决。不知道你用网络版的支持不支持。具体怎么使用 regular expression 你可以看看 站内的 老帖子。希望能帮到你。
 
回复: 紧急 求助!

For your Puzzle 2:
I used your give [pn*] [nn*] structure to get 3664 in BNC, and it seems that most of them are valid. How come you only got 25 instances?
 

附件

  • 2009-7-23 11-41-36.jpg
    2009-7-23 11-41-36.jpg
    95.3 KB · 浏览: 5
回复: 紧急 求助!

谢谢两位老师的指点!非常感谢!
还想问下,我提到的第一种情况是不是只能手动排除?
关于代词做recipient的情况我是用Corpus of contemporary American English检索的。我刚又试了,的确只出来了anyone和someone,you ,me之类的代词都没检索出来。:(
是不是我的表达式有问题?烦请各位继续指教。
 
回复: 紧急 求助!

谢谢两位老师的指点!非常感谢!
还想问下,我提到的第一种情况是不是只能手动排除? 关于代词做recipient的情况我是用Corpus of contemporary American English检索的。我刚又试了,的确只出来了anyone和someone,you ,me之类的代词都没检索出来。:( 是不是我的表达式有问题?烦请各位继续指教。

目前看来只能后期删除。这样的表达式本来就是很理想化的结构,实际语言使用时会有多种变化,不是简单的词性排列一下就能穷尽的,如:give [pn*] [nn*] 可以搜到give him advice,但是就没法搜到give him a piece of advice, give him some advices, give him some good advices, etc. 因此才需要进一步对语料做句法分析处理(parse)。

比较好用的Parsed并经过人工后纠错的语料库是UCL出的ICE-GBDCPSE

如想Parse自己的语料库,建议阅读本坛的帖子:Stanford Parser online以及相关帖子。
 
回复: 紧急 求助!

谢谢洪老师!照您的指点进行人工删除,英语的看来基本可以搞定。

我同时还需要检索每个词对应的汉语词在双宾构式中的频率。我用的是国家语委的语料库,看了表达式写法的简介依然不太明白,双宾语构式的检索表达式应该怎么写。我试了给/v(<2+!/n)和给/v(<2+!/p),好像基本没起到什么限定作用。正确的表达式应该怎么写?还请各位不吝赐教。谢谢!
 
再次求助!

上次求助后得到了大家的及时指点,非常感谢!

我同时还需要检索每个词对应的汉语词在双宾构式中的频率。我用的是国家语委的语料库,看了表达式写法的简介依然不太明白,双宾语构式(张三给了李四一本书)和介词宾语构式(张三送了一本书给李四)的检索表达式应该怎么写。我试了给/v(<2+!/n)和给/v(<2+!/p),好像基本没起到什么限定作用。正确的表达式应该怎么写?还请各位不吝赐教。谢谢!
 
Last edited:
Back
顶部