Summary Details
| Query: |
Reliability of Internet for Linguistic Purposes
|
|
| Author: | Denis V Kazakov, | |
| Submitter Email: | click here to access email | |
| Linguistic LingField(s): |
Computational Linguistics
|
|
| Summary: |
Some time ago I posted the following query on using Internet search engines for linguistic purposes: >A few weeks ago I saw somebody mentioning that he used Google to see how some English phrases are used or something. I often use Alta Vista or Yahoo in my work as a translator to find out "if it is said this way in such and such language." Does anybody have any data on the validity of such searches? Do they have any quantitative value? Any other data on the use of the Internet search engines for linguistic purposes? I wish to thank all who sent their answers, which I give below. Apparently, general-purpose search engines have greater use for languages which do not have specialized corpora (like Russian, for example). Denis Kazakov, Moscow - ------------------------------------ Dear Denis, we had a seminar on internet resources given by a group of translators from the University of Germesheim about searching tools. The Prof in charge was Prof Dr. Frank Austermuhl. He has just published a book on this topic (I can?t remember the title), but it was published by St. Jerome Publ. He gave us some hints about how to use Google or yahoo to test collocation, specially when translating into a foreign language. I hope this information can help you further. Best regards, Luciane Ferreira - ------------------------------------ Dear Denis Kazakov, It's OK to use such searches, as long as you know what you're doing. Search engines nowadays have huge coverage (> 1 bln web-pages), having index good content as well as a lot of junk. Therefore, the mere fact that you found a page through such an engine, should not be viewed as any kind of endorsement by this engine - the crawling (indexing) process is fully automatic and therefore blind. I'd suggest that whenever you find a particular phrase usage, have a look at the author and the style of writing. If the page sounds otherwise reputable, you can be moderately sure the usage is OK. Of course, pages with scientific writing would carry more weight etc. I myself used search engines in this scenario (to validate language usage), so I believe when you use your best judgement about the page source, it should be OK. Quantitative value: the more documents your search engine finds that feature a particular expression, the more confident you're in its validity. As for the engine, I'd suggest to use Google (over Altavista, Yahoo or in fact any other engine), as it's coverage and quality are unprecedented. Searching tip: if you're looking for a phrase, enclose it in quotes ("like this") for searching to get precise match (but don't use quotes if you're unsure about the phrase, and want to let the search engine find pages with the same words arranged in a possibly different order). Note also that Altavista performs stemming on its queries (i.e., "dogs" = "dog"), while Google does not. Regards, Evgeniy Gabrilovich. - ------------------------------------ Hi Denis, Just several ideas. Internet is a big corpus (there was a paper on its advantages and disadvantages: Adam Kilgariff, Web as Corpus. In Proc. of Corpus Linguistics 2001 Conference, Lancaster, UK, March-April, 2001), but for English there are more organized corpora with online access and additional possibilities, like part-of-speech tags and specification of collocations. Two examples are The Bank of English: http://titania.cobuild.collins.co.uk/form.html The British National Corpus: http://sara.natcorp.ox.ac.uk/lookup.html Their free versions are limited to 40 and 50 lines respectively, but they still affirm that "it is said this way in English". If you find anything similar for Russian or German, let me know. Best, Serge |
|
| LL Issue: | 12.2038 | |
| Date Posted: | 13-Aug-2001 | |
| Original Query: | Read original query | |
|
Back |
||
|
|
||
|
Sums main page
|
||


