I just found out, that with respect to the German language, where prefixes of verbs are separated from the verb and posed behind the object, I have no chance to find out, if a verb with a prefix is used more frequent than than the same verb without the prefix.


Er schätz sie.


Er schätzt sie ab.

Note in the latter case the infinitive form of the verb is 'abschätzen'.


cf. http://books.google.com/ngrams/info

  • 1
    The title of the question seems different than the example you gave. The title sounds like you care about words with different suffixes, but your example is for a separable prefix that could show up quite far away from the verb it is associated with. The title idea might be handled by stemming. Your example would need some other grammatical tool to analyze. – Mitch Oct 17 '11 at 14:30
  • I don't know if this helps much, but the Icelandic google (google.is) is inflection aware, so if I wanted to get a rough count of the use of say, a loan word vs the corresponding native based word, I could do a google search and see how many pages Google found and be somewhat confident that that takes into account all possible inflections of that word. This maybe true for other languages (google.de), I don't know. – MatthewMartin Oct 17 '11 at 14:53
  • 1
    @Mitch Please propose an improved title or just change it if you like. I'm interested in the problem demonstrated by the example. – bernd_k Oct 17 '11 at 15:07
  • Maybe you are looking simply for wildcards? or the appearance of two separate words? that is, look for 'schaetzt ab' and separately 'schaetzt'; the first might look for docs where both appear. And so you know you can subtract those from the second search, to get those that -don't have 'ab'. – Mitch Oct 17 '11 at 15:42
  • I'm mainly interested in engines that could be used by a web interface. I did play a bit with NLTK in the past (before I converted from Python to PowerShell), but the point is, that I'm looking for solutions, the regular user of 'German Language & Usage' can apply. There I would not expect that everyone knows about regular expressions and the like. – bernd_k Oct 17 '11 at 16:01

There is a web interface for the Verbmobil corpus that may give a rough approximation of what you need. Try the query [lemma="kommen"] []{0,10} auf for forms of the word aufkommen in which the auf is separated by up to ten words.

This is of course suboptimal; you really want to find examples in which kommen and auf are in a specific syntactic relationship. But to find examples like that, you'd need (a) a parsed corpus, and (b) a search language that can describe syntactic patterns and not just surface word order. If you wanted to go down that rabbit hole, I'd suggest the TIGER corpus, but it's not web accessible and not really usable by people who don't understand a bit about linguistics.

As Mitch points out in his comment, the separable prefixes are an unusually tough case. So comparing counts for kommen and aufkommen will actually be much more difficult than doing other sorts of comparisons. If you wanted to know something like "Is kommen (with or without a prefix) more common than gehen (with or without a prefix)," you could get a pretty good answer from Verbmobil. You need to use the advanced search so you can set the number of returned matches above 100, but if you do that, [lemma="kommen"] gives 1176 matches and [lemma="gehen"] gives 4545.

(Those counts also highlight another disadvantage of the Verbmobil corpus: it's much smaller than Google's. This is generally true in corpus linguistics: it's easy to get enormous corpora of raw text, but corpora with any sort of grammatical information attached are really labor-intensive and tend to be much smaller.)

| improve this answer | |

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.